gh-143423: Fix free-threaded build detection in sampling profiler

[DCchoudhury15]

Description

This PR corrects a logic error in Lib/profiling/sampling/sample.py where _FREE_THREADED_BUILD was incorrectly evaluated as True on Windows non-free-threaded builds when Py_GIL_DISABLED was explicitly set to 0.

The detection has been updated from an is not None check to a bool() cast, which correctly handles:

• 1 -> True
• 0 -> False
• None -> False (standard builds)

Additionally, I have updated _new_unwinder to properly handle the all_threads vs only_active_thread parameters based on the detected build type.

Tests

• Verified reproduction by checking that _FREE_THREADED_BUILD returns False when sysconfig.get_config_var("Py_GIL_DISABLED") is 0.
• Ran the official test suite: ./python -m test test_profiling (Passed).

• Issue: _FREE_THREADED_BUILD mis-detected when Py_GIL_DISABLED is defined but set to 0 #143423

[python-cla-bot]

All commit authors signed the Contributor License Agreement.

[bedevere-app]

Most changes to Python require a NEWS entry. Add one using the blurb_it web app or the blurb command-line tool.

If this change has little impact on Python users, wait for a maintainer to apply the skip news label instead.

[DCchoudhury15]

@pablogsal up for review

[vstinner]

Additionally, I have updated _new_unwinder to properly handle the all_threads vs only_active_thread parameters based on the detected build type.

I suggest reverting this change and open a separated PR for it.

[DCchoudhury15]

@vstinner I attempted to revert the changes to _new_unwinder as requested, but I found that they are actually required for the tests to pass.

If I revert to the original code, test_profiling fails because idle_worker threads are missing from the stats.

The Traceback:

test test_profiling failed -- Traceback (most recent call last):
  File "/workspaces/cpython/Lib/test/test_profiling/test_sampling_profiler/test_modes.py", line 197, in test_cpu_mode_integration_filtering
    self.assertIn("idle_worker", wall_mode_output)
AssertionError: 'idle_worker' not found in 'Captured 401 samples...

The Logic Error: The issue lies in the original code: only_active_thread=bool(self.all_threads)

When self.all_threads is True, this sets only_active_thread=True. This effectively hides idle threads, which contradicts the intention of all_threads and causes the test failure shown above.

It seems my main fix (correcting _FREE_THREADED_BUILD) unmasked this existing logic bug, as the code now enters this path where it previously didn't.

Given this dependency, is it okay to keep the _new_unwinder fix in this PR?

[vstinner]

In this case, the if self.all_threads: change should be merged first. But I still consider that it would be better to have 2 pull requests.

[DCchoudhury15]

@vstinner I have updated the PR as requested.

1. Refactored the _new_unwinder logic to use **kwargs, which removes the code duplication.
2. Reverted the _FREE_THREADED_BUILD detection fix.

This PR now focuses strictly on resolving the thread-handling logic bug. I will submit the build detection fix in a separate PR once this has been addressed.

[vstinner]

In this case, the if self.all_threads: change should be merged first. But I still consider that it would be better to have 2 pull requests.

[vstinner]

This code is almost the same as above. You should do something like:

kwargs = {}
if self.all_threads:
    kwargs['all_threads' = self.all_threads
unwinder = _remote_debugging.RemoteUnwinder(..., **kwargs)

[vstinner]

cc @pablogsal