Pyranges

- add possibility of 5-end and 3-end in slack being different/none
- add slack to join-method
- add new_position method to take union or intersection of two pairs of Start/End-columns in pyranges

- Add int64-flag to method pr.random.

- Ensure that Chromosome and Strand is str dtype before creating category
- Add check to ensure that the columns Chromosome, Start and End exist when trying to create a PyRanges

- Fix error in pypi file

Fixes:
- fix creating duplicate indexes in pyrange apply both
- fix regression where joining unstranded and stranded pyrange did not make a stranded pyrange
- default was strand=False for a few methods, should have been None (i.e. autodetect)
- read_bed now handles gzipped bed (if the file has the .gz extension)
- now able to print untraditional strands which are not strings
- fix drop when "Strand" is part of what is to be dropped
- more robust checking if column is in gr

Additions:
- print functions take formatting-argument {"Start": "{:,}"}

Changes:
- print shows sorted stranded data in Start/End order
- print dynamically selects number of untraditional strands and hidden columns to display
- read_bed now takes nrows arg
- now assertion is raised if trying to drop "Chromosome", "Start" or "End" (instead of ignoring)
- to_bed, to_gtf, to_csv now take compression argument ("infer" by default)
- to_csv writes the header as default

Additions:
- pr.itergrs added to iterate over the dfs from multiple pyranges at the same time

Changes:
- pybigwig and bamread are optional dependencies that need to be manually installed (like ray)