Pyranges

- fix gtf reading (fairliereese)
- fix FutureWarning split.py (y9c)
- add ignore_bad line to read_gtf to skip unparseable lines.

- genomicfeatures.genome_bounds can use a pyfaidx object as chromsizes
- add option genome_bounds to only clip right out-of-bounds
- genome_bounds bugfix: modified df in-place

- update numpy types
- reformat with black
- fix doctests

- fix error in null_types when a variable is pandas.StringDtype (mkanai)

- Use pd.concat instead of deprecated df.append (aallahyar)
- Keep order of non-canonical columns when writing BED (fairliereese)

marco-mariotti:
- bug fix on subsequence and spliced_subsequences, occurring when different groups of intervals overlapped, making the not contiguous in the sorted df
- expanded sort: now accepts "5" for sorting ascending on + stand, descending on - strand
- expanded gtf/gff readers
- expanded gtf/gff writer, now accepts a map_cols argument
- new method: boundaries to get boundaries of group of intervals
- expanded extend, now optionally accepts a group_by argument
- changed name of get_fasta to get_sequence
- new method: get_transcript_sequence
- docstring modified in a few places