Free-text tokens now also match Photo::people(), so a person's name typed in
Search Photos finds the photos with that named face. The search text is split
by a quote-aware tokenizer instead of split_whitespace(), so person:"Jane Doe",
who:"Jane Doe", "jane doe" and keyword:"new york" are one token each. An
unclosed quote runs to the end of the text, which can only narrow the match.
The tokenizer makes one pass and allocates nothing.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
