Implement maxpool sampling

[stephaniewankowicz]

To reduce the number of conformers needed to sample, we implement an initial round of spare sampling, maxpool the map and model and then sample more finely around the chi angles directed by maxpooling.

We tested this on all individual residues in the test dataset looking for RMSD changes > 0.5 and RSCC changes > 0.2. We only found 4 such examples.

Additionally, the Rfree stays almost exactly the same with 51% of structures having improved Rfree.

One nice feature is that we often end up modeling fewer altlocs per residue.

This will allow us to 1) speed up aromatic sampling (speeds this up by 100x) , 2) start to think about how to implement improved backbone sampling (and still allow extensive sampling in the side chain), 3) allow for this to be implemented with water sampling.

Pull Request Checklist

• Will your PR merge into the dev branch?
  Exceptions will be made for urgent bugfixes.
• Have you forked from dev?
  If not, please rebase your PR onto the most recent dev tip.
• Does your PR title succinctly describe the changes?
  Explain to a new user by completing the sentence: 'This PR will: ...'
• Fill out the template below.

---

Description of the Change

Release Notes

---