Fix: Remove stray </span> from 'archaeology' spelling normalization

[konsougiou]

What

whisper/normalizers/english.json maps the British spelling "archaeology"
to "archeology</span>" . The value contains a stray HTML closing tag.

Where

Line 89:

"archaeology": "archeology</span>",

This file is the dictionary used by EnglishSpellingNormalizer (keys =
British spelling, values = American replacement). Because of the artifact,
normalizing any text containing "archaeology" currently produces
archeology</span>, injecting a literal </span> into the output. This
affects normalized transcripts and any WER evaluation over text with that word.

Present since the initial commit (6e3be77, Sep 2022), so this affects all released versions.

Fix

Remove the trailing </span> so the value is just "archeology".

Verification

• python -c "import json; json.load(open('whisper/normalizers/english.json'))" passes (valid JSON).
• Diff is a single-line change touching only this entry.