bug#34821: discard_input_tty does not discard pending input, resulting in garbage inserted into the buffer

[Platon Pronko]

> What do you mean by "run concurrently"?  Emacs is pretty much a single
> threaded program, and there's only one Lisp thread running at any
> given time, which will execute both calls.

My knowledge of Emacs internals is pretty thin indeed. But since adding 
sleep-for results in reordering of events, I thought that there are at least 
different threads that compete for execution (concurrent, maybe not parallel). 
Also I noticed that for example read_char() calls can block for quite a lot of 
time, while Lisp code continues to run. Perhaps there is one Lisp thread and 
some other background C threads?

> You may be right, but my reasoning was that without knowing why
> there's a second buffer-switch event sometimes, we will be unable to
> devise a good solution.  IOW, I think we need to understand the issue
> better before we are ready to discuss a solution.
I agree. I'll look at what causes that second event, I'm quite interested in 
the mechanics of it anyway.