Oh boy, trying to finger out a modified amp I'm getting in a bit over my head, like you not knowing just what was done or why we can only give it a best guess. This is why I ALWAYS documented every step of a build or modification so there was NEVER any guesswork involved. Then there is The Golden Rule of Boat Anchordom, never irreversably modify a classic, for if you do its collector value drops to zero. Now you have a decision to make, whether or not to reverse all the mods based on the fact that electrical engineer(s) designed it to meet FCC certification standards, an unknown changed it. By "unknown" I mean was he an engineer or a hack? The big question is will it meet FCC spectral purity requirements or is it a Dave Made cee bee leenyaar? Hell's bells, it doesn't have any pills!

Pills = CB lingo for RF power transistors. Jokingly we called them helicopter transistors. Back to hamspeak......

That cap was an oil filled paper "bathtub" cap of the type that was common in power supplies, especially HV supplies because they never go bad under normal conditions, and they're far more compact than a 450VDC electrolytic string with associated equalizing/bleeder resistors. Johnson actually used one in the choke input HV supply in the Viking, the only transmitter they didn't cheap out on. BTW you really went over my head with ESR, I had to look up that theoretical equivalent value for inductors and caps I never had need to learn not shooting for my BSEE. (;->) C117 being part of a pi section RF filter to prevent leakage into the power supply cable and being radiated I have only a guess why that cap was in parallel with it, a lazy repair. Instead of replacing a damaged C117, just shunt it with anything in the junk box. That being the case I suggest replacing C117 with a .005uF 600V ceramic disc cap like the original. Johnson was famous for those pi section filters on everything except RF connectors, even panel meters in steel shield cans. Oh, that's the bias supply input to the 75V regulator tube, why the dropping resistor is in the supply instead of at the tube socket I have no idea.

"As it stands now in CW or LINEAR mode the voltage regulator tubes go off, which means they are
encountering an overload."
This is where I get lost reading the schematic, the eyes aren't good like they used to be, the rotary switches is where they fail me. Just one of those Mumbai Mumbles helpless desk questions, are the VR tubes all in the correct sockets?

Whoever changed the multimeter from grid and plate current to grid and cathode current royally goofed. In a grid driven RF power amplifier, cathode current is total tube current, grid, screen and plate. Often in Class C CW and plate modulated AM operating grid bias is derived from fixed bias plus rectified grid drive, at rest the fixed bias is protective bias keeping idling current down to a dull roar. Johnson engineering wired the meter to read grid and plate for a reason! <moan>

"Though for some reason, the screen current never ever registered anything at all, till this odd situation."
Screen current? You lost me there, I don't understand the metering logic at all. Ordinarily when tuning up you tune grid for maximum and adjust drive for proper grid current, then dip plate current while adjusting loading for proper plate current. Last if you use CW you should have the correct operating parameters, for SSB you adjust drive so voice peaks barely twitch grid current. That amp is modded weirdly so I wouldn't know what the readings are supposed to be now.

"Could something have changed the config back toward AB2 ?"
I think those poor 4-400s are as confused as I am.
"That oil cap perhaps?"
No, it didn't belong in there, and where it was had nothing to do with anything really.

"Also another question is, the new resistors I found (R116A-B) in the updated schematic, do they look like they are modding to AB1 operation?"
First, I have no idea why they have 3 resistors in series totaling 7.6k off the -75V regulated line. Forget what the guy who "rejiggered" it said (he sounds very unprofessional) and try to find out what Johnson did when upgrading to this model.

Changing the resting bias from -75V to -105V and even more drastic a change to -150V with a corresponding increase of screen voltage is driving them toward Class AB1. Without knowing what the TOTAL grid bias under operation is there is no way of knowing where on the curve family the tubes are operating. Your best bet is downloading a data sheet and taking measurements with the beast fired up in CW operation into a dummy load. I found that modern FETVMs are RF sensitive, older VTVMs less so, and my trusty Simpson 460 VOM was immune and at 50k per volt only loaded down AGC lines badly, a VOM would be a good instrument to measure T Bolt operating voltages. Just stay away from the 3KV without a WELL INSULATED X10 multiplier probe.

I hope I'm not confusing you TOO badly working from 40 year old memories foggy at best. Working from pdf with the + sign clicked on helped my bad eyes, but scrolling around is one PITA. (;->) BTW I doubt Covid 19 is related to tRump Derangement Syndrome, I survived the virus maybe because I never had a case of tRump anything?

RF helicopter power transistor.jpg