Can you post a picture of the strobe switch diodes in the cabinet?

That pic looks like it's under the playfield.
The diodes I'm interested in are at the bottom of the cabinet. Do you by chance have any 1N5819 diodes?

Is there a schematic available for the Ni-Wumpf board?

Well that probably explains why a capacitor across the tilt switch doesn't have the desired effect here. The capacitor is never able to charge to an open switch state because there's no diode to block it from discharging when the strobe signal releases.

JethroP try adding your capacitor on the tilt mech again, but this time add a diode - I'm guessing it needs to go to the rod side of the tilt plump bob with the banded side of the diode connected to the wire side (non banded to the rod side).

As noted in post #16 of this thread, the pull-up resistor on the switch return line the tilt switch is on is 1k ohms (Gottlieb boards have a 2.7k ohm pull-up resistor). This doesn't help a resistive plumb-bobs cause.
Is there any reason a higher value pull-up resistor (say 10k ohms) couldn't be used to make it easier for the plumb bob to pull the signal to zero volts on fast activation?

Respectfully, I don't see your logic here.
If the plumb-bob is having trouble pulling the signal LOW because of resistive contact, then a *higher* value ("weaker" as zacaj put it) pull-up resistor will allow the signal to swing low easier.

The question is whether there would be any side effects of raising the pull-up resistor value, i.e. will the R0 return signal accidentally trigger on noise interference through the cabinet/playfield wiring harness?

Out of curiosity, can you explain the Ni-Wumpf software logic to determine a "valid" tilt switch activation? i.e. how often are switches scanned, and how long do the switch return signals need to be active LOW to be considered a valid switch activation and not just noise/switch bounce? Or does your software switch matrix scan method work differently? (I come from the Bally world)

With the lower 1k ohm pull-up resistor being used, shouldn't the C7 "switch settling" capacitor be raised in value to hold an activated low signal longer?

Thanks for the explanation Ace,
I have no experience with Gottlieb in this regard as I'm from the Bally world. They require active high switch returns which seems the better approach and gives more margin to sense active TTL switch signals, but that's another matter. They use 56k ohms pull-down resistors with no switch smoothing caps on the MPU board - these caps are fitted directly and only across fast react switches (tilt mech being one of them) with a value of 0.05uF - hence the O.P asking about installing caps on the Gottlieb tilt switch plumb-bob mech.
Bally scans the switch matrix every 8.3ms. Each strobe line is active approx 257us and the switch returns are read at the 238us mark. Two consecutive active high signal returns on a switch are then considered a "valid" closure. The switch has to be serviced then sensed as open before it can be validated a new time and this works very well. I don't know if you're handling debounce in software or via the 8279 keyboard mode.

With the tilt plumb-bob closing/opening instantaneously unlike the swiping action of all other game switches seems to me makes it difficult to sense sequential closures in this scanned matrix circuit. With that in mind, if someone was to design a circuit to sense the plumb-bob activation and have it simulate a longer hard switch closure, what's the optimal time of the active low pulse you want to see?

Sorry not me, I don't have any system 1 games to test it on.

In the meantime, I still think the diode/cap has some merit, *but* a 1N4004/1N4148 diode isn't going to cut it and neither is a 0.047uf capacitor in this circuit.
The logic low level for a TTL chip is considered between 0 and 0.8 volts. The MPU board wants to see the active closed switch signal in that range for a period of time.

A low current voltage drop across a 1N4004/1N4148 diode is approx 0.72 volts leaving you next to no room and that's assuming the MPU board manages to pull the strobe line down to zero volts.
That's the reason I asked if you had a 1N5819 diode. The low current voltage drop across them is about 0.25 volts giving you more headroom before the signal upswing moves into the indeterminate logic level voltage.
The 1k ohm pull-up resistor on the MPU board will charge a 0.047uf capacitor too fast rendering it useless in this application. You might need something like a 10uF cap to hold the closed switch state longer.

But I'm only guessing - not sure I can add much more here since I don't have any games of this series to experiment with.

Disconnect the strobe wire from the tilt mech.
Banded side of the diode on the strobe wire.
Non-banded side of the diode to the tilt mech where the strobe wire was connected.
Capacitor across the two contact points of the mech.
If you're using an electrolytic capacitor, connect the negative lead to the non-banded side of the diode.
Identifying the strobe wire is the one that does NOT connect onwards to the coin switches and start button.

Hopefully this helps, if not something more fool proof needs to be done.

Great to hear! thanks for the update