Find hadiths stored under a combined number (#614)

[wakqasahmed]

Some hadiths are stored under a combined number like "6924, 6925", so /v1/collections/bukhari/hadiths/6924 and .../6925 both returned 404 because every lookup was an exact match on hadithNumber.

I followed what the website already does in Util::getURNByNumber: try the exact match first, and only if that finds nothing, do a LIKE '%<n>%' search and keep the rows where the number equals one of the trimmed parts after splitting on , or -. The split/match bit lives in a small hadith_number.py so it can be unit tested without a database (main.py can't be imported in tests because models.py reflects against MySQL). The LIKE uses a bound param with autoescape, so % or _ in the URL don't act as wildcards. I used the same fallback for GET /v1/hadiths?hadithNumber= since it's the same one-line swap. I left /v1/hadiths/refs alone because it batches all refs into a single OR query and would need a separate second pass, and I didn't touch the Muslim letter-suffix handling (#1456).

To check it, I ran the repo's docker compose stack and added one copy of Bukhari 1 to the sample DB with hadithNumber = '6924, 6925'.

Before:

GET /v1/collections/bukhari/hadiths/6924                  -> 404
GET /v1/collections/bukhari/hadiths/6925                  -> 404
GET /v1/hadiths?collection=bukhari&hadithNumber=6925      -> 200, total 0

After:

GET /v1/collections/bukhari/hadiths/6924                  -> 200, hadithNumber "6924, 6925"
GET /v1/collections/bukhari/hadiths/6925                  -> 200, hadithNumber "6924, 6925"
GET /v1/hadiths?collection=bukhari&hadithNumber=6925      -> 200, total 1
GET /v1/collections/bukhari/hadiths/692                   -> 404 (partial number, still no match)
GET /v1/collections/bukhari/hadiths/9999                  -> 404
GET /v1/collections/bukhari/hadiths/%25                   -> 404
GET /v1/collections/bukhari/hadiths/5                     -> 200, unchanged

I also added tests/test_hadith_number.py. python3 -m unittest tests.test_hadith_number tests.test_text_transform passes (10 tests), and black/flake8 are clean on the lines I changed.

Fixes #614