Fix potential deadlock when triggering project refresh while holding write lock

[embeddedt]

The common problem among all the stacktraces I looked at is a deadlock occuring when a thread holding the write lock tries to access the AWT or the write intent lock. My reading of the IntelliJ threading model documentation suggests this can be problematic. Since both of the two cases I found appear to be simple (triggering a project refresh), it seems that the simplest fix is to just queue the refresh call to be run later on the correct thread, independently of any write lock held by the surrounding method.

It will take some time to confirm if this actually fixes the problem, because it is impossible to reproduce reliably, hence why I resorted to stacktrace analysis.

In theory, this fixes #2566, fixes #2519 (there may also be other issue numbers I'm not aware of)

[Earthcomputer]

Since your other PR contained the use of LLMs (nothing wrong with that if you yourself are competent and verify yourself that it's good), could you also let me know if you used LLMs for this PR as well?

[embeddedt]

I used an LLM to point me in the direction of the locks, but ended up implementing the fix independently (its suggested solution was unnecessarily complex).