Scbs

This minor release fixes a rare ZeroDivisionError in `scbs diff`.

This minor release fixes an issue with the calculation of adjusted p-values in `scbs diff`.

This minor release addresses issues 15 and 16.

`scbs diff` now reports a lot more information and summary statistics for each DMR, including the raw p-value, the mean methylation level in both cell groups, the number of CpG sites in the DMR, the number of cells that had sequencing coverage in each group, etc.
Both `scbs diff` and `scbs scan` now optionally write a header with the `--write-header` flag.
Also `scbs filter` no longer just throws away the `log_info.txt` file and instead copies it to the filtered directory, as it should.

This release fixes 17 , now `scbs scan` reports, for each VMR, how many CpG sites the VMR contains, and how many cells have sequencing coverage in the VMR.
VMRs with low coverage, i.e. VMRs with data in very few cells, can now also be filtered automatically with the new `---min-cells` option.

This release comes with multiple new features and improvements:
* The new command `scbs diff` allows you to scan the whole genome for differentially methylated regions (DMRs) between two user-defined groups of cells. This works by sliding a window across the genome, performing a t-test for each window, and merging windows above a threshold. To control the false discovery rate, the same procedure is repeated on permutations of the data which are then used to calculate an adjusted p-value for each DMR.
* `scbs matrix` now reports the methylation matrix in a more convenient format (wide matrices instead of a huge long table).
* `scbs matrix` can now use multiple threads, which means it runs much faster when quantifying a large number of genomic intervals.
* `scbs prepare` now supports [biscuit](https://huishenlab.github.io/biscuit/) .BED files as input.

This is a performance update that lowers the amount of RAM required by `scbs prepare`. Instead of `scipy.sparse.tocsr()` we now use a custom conversion algorithm that reads each chromosome in chunks, instead of loading the whole chromosome into memory.