Fix reading PathLike objects with shp, shx and dbf args

[mwtoews]

This fixes an issue reading PathLike objects with the shp, shx and dbf args. For example:

shapefile.Reader(dbf=Path("shapefiles") / "blockgroups.dbf")

raises:

shapefile.ShapefileException: Could not load shapefile constituent file from: /path/to/pyshp/shapefiles/blockgroups.dbf

src/shapefile.py:2447: ShapefileException

Other changes:

• Use simpler PathLike alias instead of os.PathLike, as this is a highly-visible object.
• Change test_shapefile.py to use mostly Path objects.
• Change test_reader_pathlike to test_reader_path_str to test reading from a str path.
• Use tmp_pth pytest fixture to write "corrupt_too_long" file instead of modifying files in the tests directory.

[JamesParrott]

There're 4 or 5 things going on here:
i) adjust type hints in run_benchmarks.py ✅
ii) adjust type hints in shapefile.py ✅
iii) support Paths in Reader's shp, dbf and shx kwargs ✅
iv) Make 106 separate changes to the existing tests in test_shapefile.py (to use paths instead of strings), many of which
affect Reader's first positional arg (shapefile_path) instead of the shp, dbf or shx kwargs ❌
v) No new tests have been added ❌

i) ii) and iii) are nice changes. It would be good to have more pathlib.Path tests in test_shapefile.py other
than test_reader_pathlike too.

This does not mean we should ditch a single one of the existing tests that pass in strings to Reader.

[JamesParrott]

I restored the old test_shapefile.py, copied in and renamed the altered tests (from the original commit on this PR) that passed a Path to Reader as shp, dbf or shx kwargs, then rebased on master (to pick up ditching of isort).