bpo-38962: Fix reference leak in the per-subinterpreter gc by pablogsal · Pull Request #17457 · python/cpython

the-knights-who-say-ni

I'm not super convinced about this PR, but indeed it fixes the reference leaks in test__xxsubinterpreters (when all other PRs in https://bugs.python.org/issue38962 are applied first):

❯ ./python -m test test__xxsubinterpreters -R :
0:00:00 load avg: 1.10 Run tests sequentially
0:00:00 load avg: 1.10 [1/1] test__xxsubinterpreters
beginning 9 repetitions
123456789
.........

== Tests result: SUCCESS ==

1 test OK.

Total duration: 17.6 sec
Tests result: SUCCESS