I've just had a read over your experience and it's uncanny how similar it is to my intermittent HR case.
I started wondering why I had experienced Quiet Line Tests full of crackles and noise but when any internal engineer would plug the Jedec into the test socket and run tests, it would be noise free.
I went and bought an extra corded handset for running tests and I started noticing that the crackles happened more frequently on a different handset. at that point I started to wonder whether different handsets requested more current from the loop circuit than other handsets when off-hook and whether the amount of current had any bearing on the HR/semi-conductive joint (ohms law anyone?).
I then started thinking that the Jedec may pull an amount of current that effected the HR joint. If not the Jedec maybe a PQ test that sends 100 volts down the line somehow effects the HR joint and the noise goes away for a short amount of time totally eluding the engineer at the time of testing...
To try and test that theory I was wondering whether a multimeter reading voltage in parallel or connected in series to read current between different handsets would shed some light. But at that time the HR fault was discovered and fixed so it will probably remain a theory.
Just for info really H4rry ............. the testers we use at OR engineering level are JDSU or EXFO hand-Held Testers.
You are quite right with your PQT observations in that the normal PQT will remotely apply 100v as part of the scripted test, and this can 'dry out' or mask a small HR-type fault.
However, over the last 12months, our tech-heads have introduced a 'Delta R' test, which tests at a lot lower voltage (couldn't say exactly what voltage), so as not to mask the HR.
The 'Delta R' can be run as a singleton test, or on a JDSU can be run as part of a scripted test. I wouldn't know about the EXFO as I don't use one ??
The two possibilities presented on the JDSU when running a mandatory PQT are .........
1) Auto-protective PQT (This has 'Delta R' as part of the script)
2) Standard PQT.
If an EU has a high number of repeat visits it is advisable to run No.1.