When the tokeniser raises errors due to invalid byte sequences only when the parser encoding is implicit (such when parsing a file), the error locations are not properly transformed to character offsets

Linked PRs

• gh-115823: Calculate correctly error locations when dealing with implicit encodings #115824
• [3.12] gh-115823: Calculate correctly error locations when dealing with implicit encodings (GH-115824) #115949
• [3.11] gh-115823: Calculate correctly error locations when dealing with implicit encodings (GH-115824) #115950