Regarding the U and I waveforms:
It's a guessing game and rather vague, but looking at the voltage and current waveforms that were recorded can reveal hints pointing to the root cause.
The RMS recordings already showed that the voltage sag isn't caused by your facility, looking at the waveform itself might show if the voltage drop is caused by a short circuit somewhere else, a heavy load starting, a substation tap switch, ...
So you'd look at the recorded voltage waveforms at the moment the sag occurs and maybe you see one phase dropping some time before the others, or all three at the same time, do they drop to near zero and recover, is the drop edge steep or slow and then try to figure out what kind of event might cause that, and then look around if you can find such a thing.
Regarding events on a regular base:
There's a lot of tell-tales around what can cause such, including the cleaning lady pulling the server plug every friday night
Anyway, it's a good point to start and look around who might be starting e.g. a large motor or other large loads at these times. A TV transmitter station might have a backup diesel generator that is tested on a regular schedule. I don't think the crossing of 15kV line is related to your issue, anyway the local substation might be to weak, so it's kind of related to this line, but not to the fact it's running across your facility.
Running the vacuum pump motors through VFDs instead of direct connection to the mains line most probably will stop the motors grumbling and pumps spitting oil (in case the oil spitting is related to the grumbling).
Induction (asynchronous) motors don't like voltage sags, this will cause electrical phase shift in the first place, leading to jerk on the mechanical side. A VFD cannot run through the sag either, but should limit the jerking (by slowly decelerating and accelerating the motor). But I can't guarantee that this will work on your place.