False interrupts and voltage drop when touch with metal

I would have thought that it would be interesting to see how often, and under what circumstances, the interrupt fired in order to attempt to isolate the cause. At the moment, you see that only once per session so you have to restart it for each test.

Turning the lights off doesn't help. And the same happens outdoors.

This code

volatile int count = 0;

void blink(){
  digitalWrite(LED_BUILTIN, HIGH);
  count++;
}

void setup() {
  pinMode(2, INPUT_PULLUP);
  pinMode(LED_BUILTIN, OUTPUT);
  attachInterrupt(digitalPinToInterrupt(2), blink, FALLING);
  delay(5000);
  Serial.begin(2000000);
  Serial.println(count);
}

void loop(){}

prints values from 1 to 7 if I single touch the D2 pin. I guess that's just bouncing. It's much harder to trigger the interrupt by touching GND but not impossible.

There is little point because we can't duplicate your environment. Only you can do this.

Doesn't matter.

No. Most cases are insulated plastic boxes.

Are you mixing up a common connection, normally attached to the zero volt point in a circuit, and Ground a physical connection to the ground, like a stake in the ground or water or gas pipe?

What's so special about the environment? Isn't it just a breadboard?

I'm just a bit unsure about English nomenclature.

No it is where you are, the signals floating about in the air close to you. These could be florescent lights that have gone faulty, a near by transmitter, like a Taxi or one of the emergency services. It could be someone close by doing some arc welding, or even a radio ham.