Pyranges

Additions:
- pr.random(n=1000, length=100, chromsizes=None, strand=True) creates a random PyRanges from a PyRanges of chromosome sizes.

Changes:
- make __iter__ return natsorted items

Removals:
- insert. use join instead

Fixes:
- bug in boolean indexing due to __iter__ returning wrong sort order

Hotfix:
- bug in assign (strand=False, by default, not None)

Additions:
- head(n=8)
- tail(n=8)
- sample(n=8)
- set_columns(new_names) to set new column names
- argument like to drop, which takes string describing regex (gr.drop(like="_left|_right"))
- add count (number of intervals) to merge and merge_by

Fixes:
- 5X faster boolean indexing
- fix some bugs in features.introns when data was missing

Changes:
- coverage renamed to_rle
- if drop used without argument, not dropping Strand by default

hotfixes

Additions:
- cluster and merge takes argument by to only merge cluster within specific features
- gr.features.introns added. Can use by="gene" or by="transcript"
- new data: pr.data.gencode_gtf and pr.data.ucsc_bed
- can subset pyrange with boolean vector
- sort also takes argument by (sort without arg sorts on start/end)

Fixes:
- bug in subset which removed strand
- bug when setting Strand with setattr
- bug when setting Chromosome with setattr

Changes:
- new method to compute cluster (3x as fast)
- string-arg to drop not interpreted as regex
- drop or keep do not take drop_strand. Only unstrand can drop strand.

Additions:
- subsetting with new col order rearranges columns