Collapse repetitive endpoint tests with `@pytest.mark.parametrize`

[nanotaboada]

Problem

tests/test_main.py has many near-identical test functions that differ only in
an input value and an expected status code — the "returns 404" cases for GET by
id, GET by squad number, PUT, and DELETE each repeat the same
arrange/act/assert shape; the cache-header and empty-body cases likewise. This
is the xUnit "one method per case" style. pytest provides a table-driven idiom.

Proposed Solution

Group cases that share a body under @pytest.mark.parametrize:

@pytest.mark.parametrize(
    ("method", "url"),
    [
        ("get", f"{PATH}{unknown_player().id}"),
        ("get", f"{PATH}squadnumber/{nonexistent_player().squad_number}"),
        ("delete", f"{PATH}squadnumber/{unknown_player().squad_number}"),
    ],
    ids=["get_by_id", "get_by_squad_number", "delete"],
)
def test_request_unknown_resource_response_status_not_found(client, method, url):
    """Unknown resource returns 404 Not Found."""
    response = getattr(client, method)(url)
    assert response.status_code == 404

Keep cases with bespoke setup/teardown (POST/PUT that create or mutate rows) as
their own functions — parametrising those hurts readability.

Suggested Approach

1. Identify clusters with identical bodies: the 404 group, the cache-header
   pair (MISS then HIT), the 422 empty-body pair.
2. Convert each cluster to one parametrised function with explicit ids= so
   failures name the case.
3. tests/test_main.py is excluded from Black — match the existing manual
   formatting.
4. Confirm the pytest run still reports a case for every original scenario.

Acceptance Criteria

• Repetitive status-code tests consolidated with parametrize
• Each parametrised case has a readable id
• Tests with bespoke setup/teardown left standalone
• Coverage does not drop; every case still exercised
• CHANGELOG.md updated

References

• pytest — How to parametrize fixtures and test functions
• pytest — Parametrizing tests (examples)