Add custom pytest collector for BDD feature files

[lonvia]

pytest-bdd is mostly awesome but I badly miss behave's ability to show and select tests by feature file. When a BDD test under pytest-bdd fails, it is an utter pain to find the feature file with the source of the scenario.

This PR uses pytest collection hooks to manually collect feature files into pytest collection items and annotate them with the feature file name. So instead of reporting a rather meaningless:

test/bdd/test_api.py ................................................... [  6%]
................................................ss...................... [ 15%]
........................................................................ [ 25%]
........................................................................ [ 34%]
........................................................................ [ 43%]
............................                                             [ 47%]
test/bdd/test_db.py .................................................... [ 54%]
........................................................................ [ 63%]
......................s...............s................................. [ 72%]
........................................................................ [ 82%]
..s.....................s.............                                   [ 86%]
test/bdd/test_osm2pgsql.py ............................................. [ 92%]
........................................................                 [100%]

we now get

test/bdd/features/api/reverse/v1_geocodejson.feature .................      [  2%]
test/bdd/features/api/reverse/v1_geojson.feature .............              [  3%]
test/bdd/features/api/reverse/geometry.feature .....                        [  4%]
test/bdd/features/api/reverse/v1_json.feature .....................         [  7%]
test/bdd/features/api/reverse/layers.feature .................              [  9%]
test/bdd/features/api/reverse/language.feature ....                         [  9%]
test/bdd/features/api/search/language.feature .......                       [ 10%]
test/bdd/features/api/details/language.feature .......                      [ 11%]
test/bdd/features/api/details/params.feature ................               [ 13%]
test/bdd/features/api/details/simple.feature ........ss...                  [ 15%]

And when running in verbose mode, there are even line numbers:

test/bdd/features/osm2pgsql/import/entrances.feature::L4::test_import_simple_entrance PASSED                                            [  9%]
test/bdd/features/osm2pgsql/import/entrances.feature::L22::test_addresses_and_entrance_information_can_exist_on_the_same_node PASSED    [ 18%]
test/bdd/features/osm2pgsql/import/entrances.feature::L33::test_entrance_import_can_be_disabled[()] PASSED                              [ 27%]
test/bdd/features/osm2pgsql/import/entrances.feature::L33::test_entrance_import_can_be_disabled[(nil)] PASSED                           [ 36%]
test/bdd/features/osm2pgsql/import/entrances.feature::L33::test_entrance_import_can_be_disabled[{}] PASSED                              [ 45%]

You can run a single feature file using expression matching:

pytest test/bdd -k osm2pgsql/import/entrances.feature

Running a single scenario is sadly a bit less straightforward but still works:

pytest test/bdd -k 'osm2pgsql/import/entrances.feature and L4'

Only works with pytest >= 8.0. Older versions of pytest will still work but keep using pytest-bdd's default collector.