I can only speak from my own experience. I have had two touch holes burn out on liners. One was on a .36 that was used for many years and had a large enough touch hole to spray people 25 feet down the line. The other was a .32 that was my screwup when I didn't drill and tap the touch hole liner deep enough. When I filed it flush, the web was too thin and burned out on the maiden voyage.
I didn't notice too much drop off in accuracy on the .36 until the touch hole had eroded where it seemed I was getting a lot of flyers all the time and really spraying. I was concerned with how much the rifle was spraying out the touch hole more than I like. I replaced the touch hole and the problems went away.
On the .32 I was putting the first rounds through it and it was shooting well for the first few rounds, then started throwing flyers. It was also to starting to self prime. I decided to give it a few more rounds. After two shots the touch hole burned out too large to even push the ball. Everything came out the touch hole. Nothing blew but there was a long phuuut as all the pressure came out the touch hole. I took it home, pulled the ball and pulled the touch hole liner.
I found my problem was using the wrong tap. I used a normal tap on the threads where I should have used a bottom tap. The normal tap was bottoming out against the barrel wall on the little .32 and I thought I had tapped the full length on the side. I used a bottom tap and the touch hole liner went into the proper depth and the new liner has worked fine for a couple of years.
In answer to the first theory, the .36 did not shot much drop in accuracy until the touch hole got large enough to give a very noticeably large spray.
As to the second theory, the pan is open and only a small portion of the flash actually enters the barrel so contributes very little, if anything, to the charge sending the ball down the barrel. By the time the flash contribution is equal to the charge contribution the touch hole would be too large to build pressure behind the ball and go out the touch hole and join the flash.
Just some observations that might not mean much.
I would like to see some data on pressure reduction with size of touch hole though. That would be interesting. It does sound like a complex experiment though.