Possible logical flaw detected: programmers so far assumed that the same logic we implement for T230342: enable/disable save button depending on Bridge state was identical to the one to determine which bucket we count a close as, but this is not quite compatible with "even if eventually the original state is restored".

The save button be should be disabled when ever the state is not eligible for click (unchanged, invalid)
while
the tracking apparently should count as changed if the state was once considered this (irrespective of the state at the time of close).

@Pablo-WMDE Thanks! That is intended. In the second bucket we only want to know if they did anything with the data - we don't care what they did.

Thanks! That is intended

Sure. Was more meant as a note to the dev: our initial assumption ("for free") was partially flawed.

Moving this to later because we can get a basic number based on the Bridge opening stats by datatype combined with the number of edits made through the Bridge.