All fine for a week or so.
This is possibly the biggest clue.
As well as pursuing investigating a supply issue....
I'd be thinking whether anything might be out of circuit when testing that is in circuit during the fault.
Do they have any outside lights or anything time clock controlled?
e.g. an outside light switched by a double pole switch on the ring that is also linked to a PIR fed by the lighting circuit with all neutrals joined up, or something similarly special.
I'd also take a look at the boiler wiring if it's on the downstairs sockets circuit as when combi's are fitted all sorts of delightful things are done by our friends who fit them.
Have you tested the cooker itself? A cooker with a slowly developing N-E fault touching a bonded gas pipe behind can be the source of interesting fault chasing as the thing triggering it can be anything that is earthed.
I'd also try to be thinking about whether there is anywhere that the circuits run together. I remember one very obscure fault where cables had been pulled hard and they'd cut into some metal trunking, and vibrations changed the IR from absolutely fine to practically zero.
Best of luck and do report back!