Cap score set search and remove its per-result queries

[bencap]

Problem

Score set search does unbounded work per request.

• search_score_sets (routers/score_sets.py:699) caps results at SCORE_SET_SEARCH_MAX_LIMIT only when publication_identifiers is absent. A search with publication identifiers has no cap, and search_my_score_sets (:812) has none either.
• The search routes build the full score set view model for each result and then trim it to the short one, which triggers about 15 lazy-load queries per result.
• get_filter_options_for_search (:753, through fetch_score_set_search_filter_options in lib/score_sets.py) loads every matching score set plus 8–15 queries each. The ui calls it after every pause in typing, and the handler keeps running after the ui cancels.

Scope

• Apply the cap to every search, including publication-identifier searches and /me/score-sets/search.
• Build the short view model directly, with loader options for the fields it serializes.
• Compute filter options with aggregate queries over the matching set, not by loading score sets.

Acceptance criteria

• No search route returns more than SCORE_SET_SEARCH_MAX_LIMIT results.
• Query-count tests assert a constant number of queries for search and for filter options as the number of matches grows.
• Search and filter-option responses are unchanged for the existing tests.