There can be literally thousands of pins in any one bed-of-nails and that fixture, in production, might be actuated literally thousands of times in a day. I have memories of test techs banging (tappy-tap-tap) on the side of the fixture or toggling the actuation multiple times to get reliable contact for a test to commence. Often times replacing several pins to get it to work for the production batch but usually they get stuck in a retracted position, not extended which could cause a defect like that. And usually, probe marks aren't cause for rejection but that does happen to be a critical interface surface.
But we were also a low-rate, ultra-hi-reliability manufacturing facility for aerospace. Fixture designs might be somewhat different for hi-rate, just kick it out the door and let RMA pay for it when the customer catches the defect.