silently skip a test

Sometimes a test should always be skipped, e.g. because we generate tests by first generating all possible combinations of parameters and then calling pytest.skip inside the test function for combinations that don't make sense. That's different from tests that are skipped in a particular test run, but that might be run in another test run (e.g. on different hardware or when a particular feature is added). Could you add a way to mark tests that should always be skipped, and have them reported separately from tests that are sometimes skipped? When a report says "400 tests skipped" it looks like something is wrong, but maybe 390 of these should always be skipped, and just 10 should be tested later when conditions permit. Also, the "% of tests done" status message becomes distorted when always-skipped tests are included.

Start by tracing pytest.skip and the reporting of skipped tests, including the “% of tests done” status message. Define how always-skipped tests would be distinguished from skips caused by run conditions, then verify separate counts and corrected progress reporting with tests.