[Sskift]

The Z and ZZ offset tokens already accept Z and z in their regular expressions, but from_format() then tries to parse them as numeric offsets. For example, parsing 1975-05-21T22:32:11Z with YYYY-MM-DD[T]HH:mm:ssZ fails with ValueError: invalid literal for int() with base 10: ''.

Normalize the accepted UTC designators to +00:00 before the existing offset conversion. Tests cover both token forms, both designator cases, numeric-zero controls, and an explicit non-UTC default timezone.

Validation on Windows / CPython 3.12 using the pure-Python source: 8 failures and 8 passing controls before the change; all 16 pass afterward. The full suite has 1,883 passes, 5 skips and the existing timezone-name failure in test_instance_with_aware_datetime_any_tzinfo. Mypy on the changed source and all applicable pinned prek hooks pass; the Rust hook has no matching changed files. The Rust extension was not built locally.

• Added tests for changed code.
• Updated documentation for changed code. This makes the existing token grammar work as specified without adding a format token.

This contribution was prepared with Codex assistance.