Add free-threading support

[dairiki]

This is an attempt to address #460. Mostly, I followed the instructions that @AA-Turner linked to there.

I looked over _speedups.c. My non-expert eye did not see anything that didn't look nogil-safe, so I just added the Py_MOD_GIL_NOT_USED declaration to the extension module.

The matrix in the tests.yaml workflow has been expanded to included free-threading versions of python3.13 under Linux and Windows.
The Linux tests work.
The Windows tests do not, for reasons that I do not understand.

I played with enabling free-threading cibuildwheel builds in a separate branch, and that seems to just work. (Though I have not tested any of the resulting wheels to verify that they are usable.)

Fixes #460

[dairiki]

In the Windows free-threading test, pytest appears to fail with

Exception ignored on threading shutdown:
Traceback (most recent call last):
  File "C:\Python313\Lib\threading.py", line 1524, in _shutdown
    if _main_thread._handle.is_done() and _is_main_interpreter():
  py313: FAIL code 32[21](https://github.com/dairiki/markupsafe/actions/runs/10529464771/job/29177296179#step:7:22)225477 (19.61=setup[18.12]+cmd[0.03,1.46] seconds)
SystemError: <method 'is_done' of '_thread._ThreadHandle' objects> returned a result with an exception set

I have no idea what that might be about.
(I have no Windows system available on which to play around locally so further investigation is not easy.)