add temperature analysis for cluster member fraction and RMSD

[cwalker7]

Thinking about how we can use RMSD of the folded cluster as a measure of stability, and it seems like having structures from multiple temperatures in one cluster will lead to lower RMSD whenever there is a higher melting point (there are less disordered structures to discard in the density filtering step, so filtering will also get rid of noisier folded structures). So not really a fair comparison between simulations with different parameters.

Some ideas for features to add to the clustering code:

1. Compute fraction of the folded cluster structures coming from each temperature (or trajectory file in general) - I already have this mostly working.
2. Compute RMSD of subset of cluster members corresponding to each temperature - this way we can compare RMSD's with the same thermal noise.