Minimizing the stack in the last commit is meant to help with bug 54064 which I should have noted somewhere: https://bugs.winehq.org/show_bug.cgi?id=54064
I have not confirmed that it helps with bug 54064 since I don't think I can actually reproduce that failure mode here. But if it makes sense maybe it can be committed anyway and then we'll see if the failure goes away.
Also I submitted the first two patches in MR!2182 since they can be useful independently from the last commit (it's just tests and actually using the stack size set by TpSetPoolStackInformation()). https://gitlab.winehq.org/wine/wine/-/merge_requests/2182