"Are the transformers varnish coated/impregnated? If so, could the compound used slowly break down the wire covering?"
Working units all fall into a small resistance range so we are not seeing the loss of enamel leading to lower resistance of previously insulated wires now touching, as we see with U47 power transformers now.
"Does this happen with newly installed transformers or only after some time?"
I couldn't comment (no NOS inventory of the part) but perhaps others have installed NOS T14/1's which have failed. I have installed many of the clone T14/1 Oliver made for me about 13 years ago with no failures.
"Was phantom power involved with the failures?"
In my cases yes.
"What happens when you plug into a input with phantom power on?"
There was a VERY long internal AES SC0404 (the standards committe) about this, I don't recall a conclusion. Long story short there are many permutations, but the majority of gear exposed to the event is robust enough for it to be a non-event.
"I wonder if you could get a step up voltage at the cap to plate?"
This part already does not have extraordinary low frequency response so I don't see (on first glance) how it could possibly be sensitive to a step pulse from the secondary to the primary in the first place, that step is NOT getting through. If this were the cause of the failure, would we not see it in other mics with super tiny wire gauges? Example, AKG C28's using the part, not a frequent failure. Other mics using tiny Beyer or Hauffe xfmrs, though not with the same step up, do use that wire ga w/o failure.
I do recall Oliver discussing the ignition like possibilities of phantom power a few years ago, but really for all the potential damage I see little evidence that it is actually occuring (at least in ccondenser mics). I don't spend a lot of time with ribbons or dynamics so that is a different evidence pool than mine.
off topic ish I have seen the ignition phenomena destroy U47 power supply chokes.