The scope is alive and well!
The horizontal hold adjustment was extremely twitchy, and would not hold when the machine was powered off and then rebooted. I found R311, a 39K 1/4W resistor had opened, thus failing to deliver a sample of the sawtooth to the PLL in the MC1391P horizontal support chip. Since the loop couldn't lock, the horizontal oscillator was unstable. Replacing the resistor yielded a nice, stable, easily-adjusted image.
Additionally, the image was a bit narrow, but the slug in the horizontal width coil was already broken and appeared to have some sort of thread-locker on it as well, so I didn't want to fracture it further by trying to turn it. I added a little extra capacitance in parallel with the horizontal width capacitor C320 and the raster now fills the screen.
The best documentation I found for the A20 Monochrome monitor is at the end of this manual (it doesn't quite match my board 100%, but is very close):
www.tek.com/oscilloscope/tds520b-manual/tds520b-mod-cm-component-service-manualThe acrylic sheet plus a layer of silicone rubber seems to have resolved the anode-cap arcing issue.
In the end I did nothing with the splotchy anti-glare coating. But since it was noticeable by reflection more than by transmission, I reversed the glass, putting the bad face toward the CRT. There seems to be a similar, undamaged coating on the other side of the glass. The damage is really not visible.
So I am a very happy camper. Thank you so much to all who responded in this thread!