Tests should always have commentary that emphasizes why the test exists.
This is important so future refactors can know if it's reasonable to remove the test because the why is no longer valid.
Sometimes a test is straightforward enough that the name of the test can sufficiently self descriptive:
def test_create_account_prevents_duplicates_account_with_same_email(...) -> None:
"""Test that creating an account fails with expected error when the same email is used."""
...Sometimes a test exists for strange cases that would be too awkward to include in a title:
def test_batch_validation_rejects_mixed_movement_types() -> None:
"""Confirm that batch validation rejects mixed movement types"""
# legacy app Desktop exports sometimes include header rows with MovementType="H"
# mixed with actual transaction rows. The legacy import silently filtered
# these, but the new validator must explicitly reject them to avoid
# partial imports that confuse reconciliation.
...In both cases, use a docstring, but treat a docstring as a more verbose version of a title.
A comment can be used that elaborates on the special 'why'.