The only culprit was the one in unifyClock.ml::249, but
I've lazyfied most of the non-trivial verbose call.

The 2 remaining unresolved testq that were timeout-ing now pass in a few ms...

The whole non-reg test time has been divided by more than 2!