gh-101747: Fix refleak in new OrderedDict repr

[sobolevn]

After: ./python.exe -m test -v test_ordered_dict -R 3:3

Ran 279 tests in 1.460s OK . == Tests result: SUCCESS == 1 test OK. Total duration: 10.2 sec Tests result: SUCCESS 

• Issue: Refleak after new OrderedDict repr #101747

[bedevere-bot]

🤖 New build scheduled with the buildbot fleet by @corona10 for commit f8adfe9 🤖

If you want to schedule another build, you need to add the :hammer: test-with-refleak-buildbots label again.

[erlend-aasland]

The aarch64 RHEL8 refleak bot still reports a leak. Could you please examine, @sobolevn?

[serhiy-storchaka]

You can move it above the Done label and change Py_XDECREF to Py_DECREF.

[sobolevn]

Thanks, it might be a slightly more readable! I will wait for all tests to finish and apply it :)

[sobolevn]

@erlend-aasland it reports:

test_importlib leaked [6, 4, 2] references, sum=12 test_importlib leaked [4, 4, 2] memory blocks, sum=10 

I think these leaks are not related 🤔

Should I fix it here?

[erlend-aasland]

Should I fix it here?

I prefer one fix per PR; that makes it easier to bisect in case something goes wrong, and it also makes for a cleaner git history.

[sobolevn]

@erlend-aasland then I consider this done:

1. Review suggestion is applied
2. New issue is created: Refleak in test_importlib on aarch64 RHEL8 #101766

[corona10]

lgtm

[corona10]

I will merge this PR after waiting for @serhiy-storchaka approval

test_importlib leaked [6, 4, 2] references, sum=12

I can not reproduce the issue on my macOS. It sounds like we have to debug on the RHEL machine.