WIP: use HashIntoType and custom hashing instead of u64

[ctb]

Brief description:

• implement threaded consume that chunks sequences into some num chunks, each of which is processed individually, and then combined at the end;
• add criterion benchmarking;
• build new k-mer hash type HashIntoType and provide custom (no-op) hashing for HashIntoType;
• pre-allocate tables with default capacity (currently 100_000);
• replace String with &str where possible, to avoid copying/cloning;

notes for later:

fix lib when doing cargo bench:

ln -s ~/miniforge3/envs/py311/lib/libpython3.11.dylib target/release/deps/

relevant pyo3 issue:

PyO3/pyo3#4155

TODO notes:

• consumed calculation/validation?
• custom hashing fn for kmers?