gh-128889: Zero out memory ctypes for generated struct layout tests

[encukou]

GCC 15 doesn't set padding bits to zero (as it's allowed to, according to standard). Let's zero the test structs explicitly.

Even with this PR, the tests still depend on the compiler keeping padding bits unchanged when surrounding values are set. (AFAIK, this is not guaranteed by the standard, and when bitfields are involved, changing padding bits might make sense as an optimization.)
Removing that limitation will be easier with introspection attributes I want to add in gh-128715.

• Issue: test_ctypes.test_generated_structs.GeneratedTest.test_generated_data 'ManyTypesU' fails on Python 3.14 built with gcc 15 #128889

[vstinner]

LGTM: I confirm that the change does fix the #128889 issue.

Without this change, test_ctypes.test_generated_structs fails. With the change, test_ctypes.test_generated_structs pass.

[encukou]

Thank you for the check!

[bedevere-bot]

⚠️⚠️⚠️ Buildbot failure ⚠️⚠️⚠️

Hi! The buildbot ARM64 macOS 3.x has failed when building commit 13475e0.

What do you need to do:

1. Don't panic.
2. Check the buildbot page in the devguide if you don't know what the buildbots are or how they work.
3. Go to the page of the buildbot that failed (https://buildbot.python.org/#/builders/725/builds/9507) and take a look at the build logs.
4. Check if the failure is related to this commit (13475e0) or if it is a false positive.
5. If the failure is related to this commit, please, reflect that on the issue and make a new Pull Request with a fix.

You can take a look at the buildbot page here:

https://buildbot.python.org/#/builders/725/builds/9507

Failed tests:

• test.test_multiprocessing_forkserver.test_processes
• test_ssl

Failed subtests:

• test_repr_rlock - test.test_multiprocessing_forkserver.test_processes.WithProcessesTestLock.test_repr_rlock
• test_preauth_data_to_tls_server - test.test_ssl.TestPreHandshakeClose.test_preauth_data_to_tls_server

Summary of the results of the build (if available):

==

Click to see traceback logs

Traceback (most recent call last):
  File "/Users/buildbot/buildarea/3.x.pablogsal-macos-m1.macos-with-brew/build/Lib/test/_test_multiprocessing.py", line 1530, in test_repr_rlock
    self.assertEqual('<RLock(SomeOtherThread, nonzero)>', repr(lock))
    ~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
AssertionError: '<RLock(SomeOtherThread, nonzero)>' != '<RLock(None, 0)>'
- <RLock(SomeOtherThread, nonzero)>
+ <RLock(None, 0)>


Traceback (most recent call last):
  File "/Users/buildbot/buildarea/3.x.pablogsal-macos-m1.macos-with-brew/build/Lib/test/test_ssl.py", line 5094, in test_preauth_data_to_tls_server
    self.assertIn("before TLS handshake with data", wrap_error.args[1])
    ~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
AssertionError: 'before TLS handshake with data' not found in '[SSL: RECORD_LAYER_FAILURE] record layer failure (_ssl.c:1066)'

[befeleme]

I can confirm that the tests executed during the rpmbuild pass with the fix.