refactor: share message text constraints between send and edit requests

[lesnik512]

Closes #18

What changed

• app/schemas/api.py: new module-level alias MessageText = Annotated[str, pydantic.Field(min_length=1, max_length=4000)]. SendMessageRequest.text and EditMessageRequest.text both use it, so the bounds 1 and 4000 are declared once in app/.
• tests/test_schemas.py: tests parametrised over both request models. Each model rejects "" and a 4001-character string and accepts a 4000-character string. If the two models drift apart later, one parametrised case fails.

Why

The two request models declared the same pydantic.Field(min_length=1, max_length=4000) independently. Changing one would have left the other alone, and no test would have noticed.

A plain Annotated assignment is used instead of a PEP 695 type alias. That keeps the field inlined in the generated OpenAPI schema exactly as before.

Test-first evidence

This is a pure refactor, so a correct test cannot fail first. The new tests were written and run against the unchanged code: 7 passed. To check that the parametrisation catches drift, I temporarily changed only EditMessageRequest.text to max_length=5000 and reran: FAILED tests/test_schemas.py::test_message_request_rejects_out_of_bounds_text[over_max-EditMessageRequest] (DID NOT RAISE ValidationError), 1 failed and 6 passed. Then I reverted it and applied the refactor: 7 passed.

Verification

• just test: 116 passed, coverage 100.00%.
• just lint: ruff format, ruff check and ty check all clean.
• OpenAPI: a Litestar app built with both models generates {'type': 'string', 'maxLength': 4000, 'minLength': 1} for text on both SendMessageRequest and EditMessageRequest, the same as before.
• Existing send/edit API tests are unchanged and pass.