Problem

DocumentTypeRouter compiles each declared MIME type as a regex and matches with
pattern.fullmatch(mime_type). A standard IANA type that contains a regex
metacharacter is therefore misinterpreted. The clearest case is the + in
image/svg+xml: as a regex, svg+xml means sv followed by one-or-more g
followed by xml, so an actual image/svg+xml document never matches and lands
in unclassified.

router = DocumentTypeRouter(mime_type_meta_field="mime_type", mime_types=["image/svg+xml"])
router.run(documents=[Document(content="<svg/>", meta={"mime_type": "image/svg+xml"})])
# -> {"unclassified": [...]}   (expected: {"image/svg+xml": [...]})

Fix

Match declared MIME types by exact equality first, then fall back to regex.
This routes literal types with metacharacters correctly while preserving regex
patterns such as audio/.*. The output bucket key is unchanged (it equals the
declared string, as before), so declared output types are unaffected. This
mirrors how FileTypeRouter handles the same situation.

Tests

Added test_run_with_literal_mime_type_containing_regex_metacharacter in
test/components/routers/test_document_type_router.py, asserting image/svg+xml
routes to its own bucket and a coexisting audio/.* regex pattern still
matches. Fails before the change, passes after. All 19 router tests pass.

$ python -m pytest test/components/routers/test_document_type_router.py
19 passed

ruff check / ruff format --check pass. Added a release note.

Disclosure: I used AI assistance (Claude) to find this regex-metacharacter
routing bug and draft the test. I reviewed the change, ran the suite, and take
responsibility for its correctness.