Cnv-from-bam

The Beta releases for 0.4.4 have more detail, but in essence, this adds bug fixes for:

* Coordinates for returned bins based off of bin width by rounding it the nearest `bin_size` (default 1000), stopping us from rounding down and having a cumulative drift of the remainder
* Low coverage read numbers return a NaN when trying to divide by 0 (empty read mapping start bins)
* `iterate_bam_file` can accept `None` as file path now
* The CI building for maturin action cross for linux `aarch64` has `gcc 4.8.5`, but `libdeflate` just dropped support for `gcc <4.9`. `libdeflate` is a dependency of `noodles-bgzf`, so we had to switch to `manylinux-2_28:cross`, which has `gcc 7`. Means we won't support CentOS 7, but I'm like 70% sure it's EOL anyway and if it isn't I don't care?

And adds the feature of:
* kwargs to the python function ( documented in README.md) which can set some of the parameters used in the calculation of the CNV.

What's Changed
* Add kwarg for fixed bin width, ploidy, bin size, number of reads in bin by Adoni5 in https://github.com/Adoni5/cnv_from_bam/pull/20

**Full Changelog**: https://github.com/Adoni5/cnv_from_bam/compare/v0.4.3...v0.4.4

Fixes 21 - which helpfully contains no information about the issue.

Basically by not rounding the calculated CNV bin_width, we were duplication the difference between the nearest 1000 (rounded up) and the acual base number for each bin, resulting in contigs which were too long when plotted via

python
bin_width * number_of_bins


So this has been fixed by rounding the CNV bin_width to the nearest 1000 and using that throughout the calculations.

Adds documentation for the keyword arguments as well.

If confirmed working, will be ready for merging the PR and full release.

Needs the changelog updating!

**Full Changelog**: https://github.com/Adoni5/cnv_from_bam/compare/v0.4.4-beta.1...v0.4.4-beta.2

Allows the `bam_path` to be set to `None`, as long as a dictionary of previously binned mapping starts are provided.
If the `bin_width` kwarg is altered will return the new CNV.

**Full Changelog**: https://github.com/Adoni5/cnv_from_bam/compare/v0.4.4-beta.0...v0.4.4-beta.1

Closes 19

Adds kwargs to `iterate_bam_file`. Possible values are `bin_width`, `bin_size`, `ploidy` and `target_reads`.
Fixes these values for the `CNV` calculations.

This is a Beta release, for testing before being merged into main and fully released.

* It is possible to iteratively add bam files to a continuing count. By passing a dictionary to iterate_bam_file, the intermediate mapping start counts are kept in this dictionary.
This is limited to parsing files one at a time, rather than by directory.
See example above under [iterative use.](iterative-use)

* Catches bug where metadata (mapped and unmapped count) for a reference sequence in a BAI or CSI file would return None, and crash the calculation. As this was used to calculate the Progress bar total, just skips the offending reference sequence, returning - for both counts. May mean the progress bar can have a lower total than number of reads, but won't matter to final numbers.

* Adds Cargo tests to CI

What's Changed
* Feature/iterative update by Adoni5 in https://github.com/Adoni5/cnv_from_bam/pull/18


**Full Changelog**: https://github.com/Adoni5/cnv_from_bam/compare/v0.4.2...v0.4.3

Highlights
* Returns the contig names naturally sorted, rather than in random order!! For example `chr1, chr2, chr3...chr22,chrM,chrX,chrY`!
Huge, will prevent some people getting repeatedly confused about expected CNV vs. Visualised and wasting an hour debugging a non existing issue.
* Returns variance across the whole genome in the CNV result struct.

What's Changed
* Feature/variance by Adoni5 in https://github.com/Adoni5/cnv_from_bam/pull/16


**Full Changelog**: https://github.com/Adoni5/cnv_from_bam/compare/v0.4.1...v0.4.2