Get the "Current" Thread State from a Thread-local

(This is a revival of gh-84702, which @vstinner worked on several years ago.)

For a per-interpreter GIL (PEP 684) we need to start storing the "current" thread state in a thread-local variable, rather than _PyRuntime.tstate_current. There may be other benefits to the approach if we can take advantage of native compiler support (e.g. C11's __thread_local).

Also see https://discuss.python.org/t/how-to-get-the-current-thread-state/22655.

Linked PRs

• gh-103323: Get the "Current" Thread State from a Thread-Local Variable #103324
• gh-103323: Remove PyRuntimeState_GetThreadState() #104171
• gh-103323: Remove current_fast_get() unused parameter #114593