PyUnicode_DecodeUTF8Stateful() should save the number of successfully decoded bytes in *consumed. But if all bytes are in the ASCII range, it uses a fast path and does not set *consumed.

It was found during writing coverage tests for Unicode C API (#99593).

Linked PRs

• gh-99612: Fix PyUnicode_DecodeUTF8Stateful() for ASCII-only data #99613
• [3.11] gh-99612: Fix PyUnicode_DecodeUTF8Stateful() for ASCII-only data (GH-99613) #107224
• [3.10] gh-99612: Fix PyUnicode_DecodeUTF8Stateful() for ASCII-only data (GH-99613) (GH-107224) #107230
• [3.9] gh-99612: Fix PyUnicode_DecodeUTF8Stateful() for ASCII-only data (GH-99613) (GH-107224) #107231