Good news! I managed to get it to fail again. Quite interesting series of events.

Tried a number of times sitting in the driveway and it always started. Took it for a 10 minute drive, and no problems. Back in the driveway, and before even attempting to start again, I checked for any new error codes - I get the following new ones: https://drive.google.com/file/d/1gHr...ew?usp=sharing

F6 IKE: No CAN identification (DME)
F5 IKE: No CAN identification (ASC)
F7 IKE: No CAN identification (DME)
5140 No message, receiver EGS, transmitter DME-DDE

So these messages aren't just due to 'low voltage' cases from cranking too much. At this point, I'm like - I bet it won't start now.

But what do you know it starts... I turn it off, and try to start again (can clearly here fuel pump running in both cases), and while still monitoring rail pressure in ISTA (in the same recording), it fails to start: https://drive.google.com/file/d/1LmI...ew?usp=sharing

(Fail start attempt begins ~38 seconds in, successful start at beginning, some more fail attemps afterwards with longer cranking)

I see the same behaviour I mentioned previously, RPM goes up with cranking, rail pressure gets to about 70 BAR, then decreases while still cranking. At some point I loose connection to the DDE with ISTA when I try to start again. To reiterate, I can definitely hear the inline fuel pump running even when it fails to start.

After these start fails, I have some extra new error codes to (as well as the previous ones): https://drive.google.com/file/d/1chZ...ew?usp=sharing

D35C DSC: No message (transfer case 192)
5141 No Message, receiver EGS, transmitter DSC
4B90 DDE: Rail-pressure monitoring on engine start
53FF VTG: Supply, control unit: terminal 15

So definitely some electric shenannigans going on. I'll go pull and replace the relays in the ebox under the bonnet again and see if that fixes things. The other possibility is the ignition switch as suggested earlier, but I don't have any other the other symptoms (dash lights doing weird stuff).

Edit: So I didn't need to pull any relays to fix it, simply clearing all the codes and it started with no issues a number of times. So it seems possibly while driving with higher vibration level or something some communication issues start happening.