Use bioframe inside skgenome

[etal]

See: https://github.com/open2c/bioframe

Why:

• Improve performance and robustness of the genomic interval arithmetic functions, especially in WGS mode.
• Reduce the maintenance surface area inside CNVkit's codebase.

There are potentially faster options, but:

• pyranges_1.x with ruranges -- evolving fast, too fast, and it brings a Rust dependency.
• polars_bio -- much heavier dependency chain, overkill here.
• scikit-bio -- interval tree is not the native data structure, so it's neither lighter nor faster than bioframe for CNVkit's use case.

Supersedes #476.