gh-159071: Correct UnicodeDecodeError offsets in utf-8-sig codec when input starts with BOM

[devpryan7792]

When the input starts with a UTF-8 BOM (codecs.BOM_UTF8), encodings.utf_8_sig was slicing the input (input = input[3:]) before calling codecs.utf_8_decode. As a result, any raised UnicodeDecodeError reported offsets and object attributes relative to the sliced input without the BOM (i.e. start was off by 3, and object omitted the initial BOM).

This aligns utf-8-sig with other BOM-stripping codecs (utf-16 and utf-32), which preserve the BOM in the exception's object, start, and end attributes:

1. Updated decode(), IncrementalDecoder._buffer_decode(), and StreamReader.decode() in Lib/encodings/utf_8_sig.py to pass the original input to codecs.utf_8_decode and strip the leading \ufeff from the decoded output.
2. Added unit test test_decode_error_offsets_with_bom to UTF8SigTest in Lib/test/test_codecs.py.
3. Added NEWS blurb entry.

Fixes #159071.

• Issue: utf-8-sig: UnicodeDecodeError offsets do not count the BOM #159071

[python-cla-bot]

All commit authors signed the Contributor License Agreement.