gh-101277: Isolate itertools, part 3/4

[erlend-aasland]

Add accumulate, compress, filterfalse, count, pairwise, zip_longest, and product types to module state

• Issue: Isolate the itertools extension module #101277

[bedevere-bot]

🤖 New build scheduled with the buildbot fleet by @erlend-aasland for commit a02e90a 🤖

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

[erlend-aasland]

@kumaraditya303: we're failing test_finalize_structseq of test_embed. the type version contract is invalidated¹ and the Py_TPFLAGS_VALID_VERSION_TAG assert in _PyType_Lookup is triggered here:

cpython/Objects/typeobject.c

Lines 4172 to 4178 in c4de6b1

	if (entry->version == type->tp_version_tag &&
	entry->name == name) {
	assert(_PyType_HasFeature(type, Py_TPFLAGS_VALID_VERSION_TAG));
	OBJECT_STAT_INC_COND(type_cache_hits, !is_dunder_name(name));
	OBJECT_STAT_INC_COND(type_cache_dunder_hits, is_dunder_name(name));
	return entry->value;
	}

AFAICS, the contract between tp_version_tag and Py_TPFLAGS_VALID_VERSION_TAG seems to hold all across the APIs² implemented typeobject.c, and from what I can see, the runtime uses only these APIs to mess with the type version.

Any ideas?

Footnotes

1. tp_version_tag is set, but tp_flags & Py_TPFLAGS_VALID_VERSION_TAG is false ↩

2. PyType_Modified, assign_version_tag, etc. ↩

[kumaraditya303]

Hmm, debugging shows that it is sys.int_info for which it is failing, this PR has nothing to with that so I guess it is a runtime bug somewhere.

Note for self:

Loop #2 ---\nobject address  : 0x5633734aa7e0\nobject refcount : 30\nobject type     : 0x563373352ce0\nobject type name: type\nobject repr     : <class 'sys.int_info'>\n_testembed: Objects/typeobject.c:4176: _PyType_Lookup: Assertion `_PyType_HasFeature(type, Py_TPFLAGS_VALID_VERSION_TAG)'

[erlend-aasland]

Hmm, debugging shows that it is sys.int_info for which it is failing, this PR has nothing to with that so I guess it is a runtime bug somewhere.

Hm, but it does not fail on main; it only fails with this patch applied. I've tried (locally) to pull in latest main, just to be sure.

[erlend-aasland]

FTR, Kumar found the culprit. See #101696.

[erlend-aasland]

Thanks, Kumar! Let's land.