gh-109162: libregrtest: move code around

[vstinner]

• Move Regrtest.display_header() to utils.py.
• Move cleanup_temp_dir() to utils.py.
• Move list_cases() to findtests.py.

• Issue: Refactor test.libregrtest #109162

[vstinner]

Strange, test_regrtest.test_huntrleaks() failed on Windows x64, but only test_huntrleaks(): test_huntrleaks_mp() passed for example whereas it's basically the same test run in a worker process.

• expected: test_regrtest_huntrleaks leaked [1, 1, 1] references, sum=3
• got: test_regrtest_huntrleaks leaked [1, 1, 9] references, sum=11

I don't know where these +8 references are coming from.

I suppose that it's random failures. Maybe related to an internal Python cache.

FAIL: test_huntrleaks (test.test_regrtest.ArgsTestCase.test_huntrleaks)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "D:\a\cpython\cpython\Lib\test\test_regrtest.py", line 1060, in test_huntrleaks
    self.check_huntrleaks(multiprocessing=False)
  File "D:\a\cpython\cpython\Lib\test\test_regrtest.py", line 1057, in check_huntrleaks
    self.check_leak(code, 'references', multiprocessing=multiprocessing)
  File "D:\a\cpython\cpython\Lib\test\test_regrtest.py", line 1039, in check_leak
    self.assertIn(line2, output)
AssertionError: 'test_regrtest_huntrleaks leaked [1, 1, 1] references, sum=3\n' not found in '0:00:00 Run tests sequentially\n0:00:00 [1/1] test_regrtest_huntrleaks\nbeginning 9 repetitions\n123456789\n.........\ntest_regrtest_huntrleaks leaked [1, 1, 9] references, sum=11\ntest_regrtest_huntrleaks leaked [1, 1, 1] memory blocks, sum=3\ntest_regrtest_huntrleaks failed (reference leak)\n\n== Tests result: FAILURE ==\n\n1 test failed:\n    test_regrtest_huntrleaks\n\nTotal duration: 496 ms\nTotal tests: run=1\nTotal test files: run=1/1 failed=1\nResult: FAILURE\n'

[vstinner]

Strange, test_regrtest.test_huntrleaks() failed on Windows x64

I cannot reproduce the issue on Windows or Linux with ./python -m test -v test_regrtest -j1 --forever. The test passed 3-4 times on each OS (I didn't wait longer).

[vstinner]

I rebased my PR on main to get "gh-107219: Fix concurrent.futures terminate_broken() (#109244)" fix. test_concurrent_futures hanged twice on the Windows x64 job.