Speed up kmer indexing

[Adamtaranto]

Kmer indexing currently uses python to loop over sequences extract each kmer and add index position to a dict. See find_match_pos_diag()

This is not practical for chromosome scale sequences due to long compute time. The kmer indexing and location lookup task could be sped up by writing this function in Rust (i.e. using Rust-Bio FM-Index).

Would require adding Maturin as a build dep for mixed rust / python project.

[Adamtaranto]

In the meantime, we could save a lot of time by pre-calulating kmer indicies for each seq outside the find_match_pos_diag() function.

Atm that function re-counts kmers every time a sequence is compared to something else. Time scales quadratically for kmer indexing and diag finding.

If we recycle the kmer index for each sequence we can make the indexing step scale linearly, this would save time for plots with lots of comparisons.