Optimize runtime and memory usage

[jranek]

What is the purpose of this PR?

Update repo to improve runtime and memory usage.

How did you implement your changes

• vectorize kh operations in kernel_herding
• reduce memory usage of phi computation in random_feats
• update rng to better handle parallelization while preserving legacy seed output in random_feats
• change parallelization to threading and remove defunct progress bar in sketch
• parse different sparse input data formats in _parse_input
• update tests in .github/workflows/build.yml

Remaining issues

NA