To follow up. It was hardware: broken vcc pin on the dht11 corrupted the whole sensor code badly. Resolder and added wait-code, inverted the on-off 0-vs-1 and I have a good working sensor. The duplicate is also installed and working reliably.
A 6-relay board is also working nicely and Domoticz is great. My old system is nearly replaced and I couldnt be happier.

Respect and thanks all around!

@Yveaux and @boum You two beat me by 2hrs.
@user2684 whose code this is (cudos to him, have used this bit on a number of sensors!) had probably figured that out and I screwed myself when I changed the original analogRead to int.

Now the whole thing works 100%

@gohan N00b question, but which steps from the step-by-step need to be redone to "compile it" after altering myconfig.h? Or can you just post the command lines?

I have now. I always have that nagging feeling...
But after MANY iterations, finally success. At least partly.

BlinkWithoutDelay works AND mysensors sketch as well (will do more testing...).
What made it work, if anyone else struggles with this:

-connect uno (to computer-linux), board:uno, upload ArduinoISP, disconnect
-add shield to uno and atmega328p-pu to shield, connect to computer, board: "8mhz on breadboard", "programmer: "Arduino as ISP", burn bootloader, disconnect
-board: Uno, connect as above, "upload sketch using programmer", disconnect EDIT this last part appears to work but internal clock is all messed up. Bootloader works fine like this.

The magic happened after I changed the board back to Uno after burning bootloader. Why is beyond me, but maybe some arduino-ninja can figure it out.

Nevermind. Whined too soon.
Completely worked over the node4 code, moved "include"s around and eventually changed the relay presentations and receive clauses. Hope this helps someone with their problems later:
Thanks a millions!

void before()
{
    // Set relay pin in output mode
    pinMode(RELAY_1, OUTPUT);
    // Set relay to last known state (using eeprom storage)
    digitalWrite(RELAY_1, loadState(CHILD_ID_RELAY)?RELAY_ON:RELAY_OFF);
}

later this....

present(CHILD_ID_RELAY, S_BINARY);


....and ending with this...

void receive(const MyMessage &message)
{
if (message.type==V_STATUS) {
    if (message.destination == MY_NODE_ID) { // Only switch the relay if the message is for this node
        Serial.print("Updating relay state");
        // Change relay state
        digitalWrite(RELAY_1, message.getBool()?RELAY_ON:RELAY_OFF);
        // Store state in eeprom
        saveState(CHILD_ID_RELAY, message.getBool());
        // Write some debug info
    }
    // Write some debug info
    Serial.print("Incoming change for sensor:");
    Serial.print(message.sensor);
    Serial.print(", New status: ");
    Serial.println(message.getBool());
}
 
}

@anticimex I've built a new node, DHT11 that sends temp&hum data. I reconfigured Rpi with this:

sudo ./configure --my-transport=nrf24 --my-rf24-irq-pin=15 --my-signing-debug --my-signing=password --my-signing-password=XXXXXX --my-gateway=mqtt --my-controller-ip-address=127.0.0.1 --my-mqtt-user=YYYY --my-mqtt-password=ZZZZZ --my-mqtt-publish-topic-prefix=mysensors-out --my-mqtt-subscribe-topic-prefix=mysensors-in --my-mqtt-client-id=mygateway1 --my-leds-err-pin=12 --my-leds-rx-pin=16 --my-leds-tx-pin=18

This is what I get in gw syslog:

Apr  2 22:53:11 DietPi mysgw: Starting gateway...
Apr  2 22:53:11 DietPi mysgw: Protocol version - 2.2.0
Apr  2 22:53:11 DietPi mysgw: MCO:BGN:INIT GW,CP=RNNGLSQX,VER=2.2.0
Apr  2 22:53:11 DietPi mysgw: !SGN:BND:PWD<8
Apr  2 22:53:11 DietPi mysgw: !SGN:INI:BND FAIL
Apr  2 22:53:11 DietPi mysgw: TSF:LRT:OK
Apr  2 22:53:11 DietPi mysgw: TSM:INIT
Apr  2 22:53:11 DietPi mysgw: TSF:WUR:MS=0
Apr  2 22:53:11 DietPi mysgw: TSM:INIT:TSP OK
Apr  2 22:53:11 DietPi mysgw: TSM:INIT:GW MODE
Apr  2 22:53:11 DietPi mysgw: TSM:READY:ID=0,PAR=0,DIS=0
Apr  2 22:53:11 DietPi mysgw: MCO:REG:NOT NEEDED
Apr  2 22:53:11 DietPi mysgw: MCO:BGN:STP
Apr  2 22:53:11 DietPi mysgw: MCO:BGN:INIT OK,TSP=1
Apr  2 22:53:11 DietPi mysgw: GWT:RMQ:MQTT RECONNECT
Apr  2 22:53:11 DietPi mysgw: connected to 127.0.0.1
Apr  2 22:53:11 DietPi mysgw: GWT:RMQ:MQTT CONNECTED
Apr  2 22:53:11 DietPi mysgw: GWT:TPS:TOPIC=mysensors-out/0/255/0/0/18,MSG SENT
Apr  2 22:53:23 DietPi mysgw: TSF:MSG:READ,99-99-255,s=255,c=3,t=7,pt=0,l=0,sg=0:
Apr  2 22:53:23 DietPi mysgw: TSF:MSG:BC
Apr  2 22:53:23 DietPi mysgw: TSF:MSG:FPAR REQ,ID=99
Apr  2 22:53:23 DietPi mysgw: TSF:PNG:SEND,TO=0
Apr  2 22:53:23 DietPi mysgw: TSF:CKU:OK
Apr  2 22:53:23 DietPi mysgw: TSF:MSG:GWL OK
Apr  2 22:53:23 DietPi mysgw: SGN:SKP:MSG CMD=3,TYPE=8
Apr  2 22:53:23 DietPi mysgw: TSF:MSG:SEND,0-0-99-99,s=255,c=3,t=8,pt=1,l=1,sg=0,ft=0,st=OK:0
Apr  2 22:53:25 DietPi mysgw: TSF:MSG:READ,99-99-0,s=255,c=3,t=24,pt=1,l=1,sg=0:1
Apr  2 22:53:25 DietPi mysgw: SGN:SKP:MSG CMD=3,TYPE=24
Apr  2 22:53:25 DietPi mysgw: TSF:MSG:PINGED,ID=99,HP=1
Apr  2 22:53:25 DietPi mysgw: SGN:SKP:MSG CMD=3,TYPE=25
Apr  2 22:53:25 DietPi mysgw: TSF:MSG:SEND,0-0-99-99,s=255,c=3,t=25,pt=1,l=1,sg=0,ft=0,st=OK:1
Apr  2 22:54:18 DietPi mysgw: TSF:MSG:READ,99-99-0,s=1,c=3,t=16,pt=0,l=0,sg=0:
Apr  2 22:54:18 DietPi mysgw: SGN:SKP:MSG CMD=3,TYPE=16
Apr  2 22:54:18 DietPi mysgw: !SGN:NCE:GEN
Apr  2 22:54:23 DietPi mysgw: TSF:MSG:READ,99-99-0,s=0,c=3,t=16,pt=0,l=0,sg=0:
Apr  2 22:54:23 DietPi mysgw: SGN:SKP:MSG CMD=3,TYPE=16
Apr  2 22:54:23 DietPi mysgw: !SGN:NCE:GEN

This is kicking my butt... I cant understand that last part about the nonce
I will add the code from the node as soon as possible.

Just a quick idea: are you powering the RFM69 from the Rpi?
I use a Rpi Zero W and when I powered my radio NRF24 from that it doesnt get enough juice even with caps.
I now use a bigger USB-wall wart and use AMS1117-3.3 to power the radio (PA-version) and added caps.

My two cents

@mfalkvidd Made the password 10 characters. From looks of the logs, it's working now.
I cant believe I skipped the step of making the password longer... Just got too fixated on the password I came up with.

Apparently sleeping less and debugging late solves problems. This is SOLVED.

My IDE library was outdated. I updated my Arduino Mysensors-library, upload to one node, and my node started to communicate fine. Must reupload all nodes unfortunately. Probably the problems are caused by signing differences in previous library versions vs. 2.3.0 release for Raspberry.

Again, hope this helps someone who encounters the same problem.

@ej3-martin My LEDs come on when the Rpi boots (is that what you mean by testing?). Also you can invert the LEDs (on but blink off) if you want. I figured the wiring makes for better power delivery rather than feeding voltage from the GPIO

I found this old thread and since I have a Vallox machine it's interesting. RS485 is new to me but MySensors isn't.

Can I have a node like this described by Heinz parallel with the "mickey mouse"-display? Using the display's connector? Or is it one or the other?

I'd feel better having both (manual display&buttons and connection to MySensors=>Domoticz, in my case).

@ej3-martin My LEDs come on when the Rpi boots (is that what you mean by testing?). Also you can invert the LEDs (on but blink off) if you want. I figured the wiring makes for better power delivery rather than feeding voltage from the GPIO

@Nicolás-Potier that's way way over my understanding and fluency with Mysensors. And way off topic, I think.

You're not being able to connect to your MQTT broker (Mosquitto or Moqutte as was in mycontroller.org). I havent used mycontroller so I cant help with that.

I use Mosquitto on the same Rpi (so just install mosquitto and mosquitto-clients). You should secure it but test first without credentials to get going. Remember to create credentials in mosquitto AND reconfig your gw

@Yveaux and @boum You two beat me by 2hrs.
@user2684 whose code this is (cudos to him, have used this bit on a number of sensors!) had probably figured that out and I screwed myself when I changed the original analogRead to int.

Now the whole thing works 100%

I dont know what controller you are using but I had similar problems with domoticz when presenting s_light_level and V_Level vs v_light_level. It had to do with domoticz update.

Thats what came to mind. I didnt find a sketch problem if this wasn't part of your problem.

@electrik Thanks I'll check those.
I added a lot of serial.Print's to debug where it goes south and it appears to ruin already at the first division after the analogRead (I started receiving 530's analogRead at room temp). So probably something else completely that's rotten. I'll try another machine with another Arduino IDE. It should work now that the original question is answered.
So I'll consider this solved. Thanks again!

@electrik ..and then just adjust the volts/bit for 3.3 rather than 1.1v, correct? (i.e x3). That fixed it, thank you.
BUT presented a weird problem. The analogRead gives a proper value but the math then is not working and I cant figure it out. I did the calculations myself and get a proper result:

int _nominal_resistor2 = 10000;
  int _nominal_temperature2 = 25;
  int _b_coefficient2 = 3950;
  int _series_resistor2 = 10000;
  int adcTmm = analogRead(THMR_PIN);
  Serial.println(adcTmm);
  float readingSau = (1023 / adcTmm)  - 1;
  readingSau = _series_resistor2 / readingSau;
  float temperatureS = readingSau / _nominal_resistor2;     // (R/Ro)
  temperatureS = log(temperatureS);                  // ln(R/Ro)
  temperatureS /= _b_coefficient2;                   // 1/B * ln(R/Ro)
  temperatureS += 1.0 / (_nominal_temperature2 + 273.15); // + (1/To)
  temperatureS = 1.0 / temperatureS;                 // Invert
  temperatureS -= 273.15;                         // convert to C
  send(msgTempSau.set((int16_t)temperatureS, 2));
    Serial.print("Sauna T: ");
    Serial.print(temperatureS);
    Serial.println(" C");

But still the serial monitor shows:

7409 MCO:SLP:WUP=1
7413 TSF:TRI:TSB
Sauna door: 1
7430 TSF:MSG:SEND,21-21-10-0,s=3,c=1,t=16,pt=0,l=1,sg=0,ft=0,st=OK:1
7966 TSF:MSG:SEND,21-21-10-0,s=1,c=1,t=0,pt=7,l=5,sg=0,ft=0,st=OK:29.00
Shower T: 29.00
7989 TSF:MSG:SEND,21-21-10-0,s=0,c=1,t=1,pt=7,l=5,sg=0,ft=0,st=OK:87.0
Shower H: 87.00
536
8005 TSF:MSG:SEND,21-21-10-0,s=16,c=1,t=0,pt=7,l=5,sg=0,ft=0,st=OK:-273.00
Sauna T: -273.15 C
Sauna off, sleeping 30min
Battery Voltage: 4.22 V   Battery percent: 97.70 %
8028 TSF:MSG:SEND,21-21-10-0,s=255,c=3,t=0,pt=1,l=1,sg=0,ft=0,st=OK:97
8097 MCO:SLP:MS=1800000,SMS=0,I1=1,M1=1,I2=255,M2=255
8108 TSF:TDI:TSL

I changed variable names to account there wouldnt be overlap with DHT library & dht part of sketch.

I investigated the bootloader problem, I use 1.8.1 definitions that made those problems disappear. But still, the A2 (THMR_PIN) reads 1023 but the battery sense on A4 works.

EDIT: now I get what you mean: it also changes the reference for the whole ADC...from 3.3 to 1.1V! Any ideas for changing the resistors? I would want to use the same thermistor as I only have two wires going to the sauna. OR is there a way to change the reference for A2 read and go back after it's been read?

Same result float => int change in the sketch.

Tried to reupload bootloader but got a problem, same with another atmega328. Will investigate that next.