ML classifier notebook cleanup and add new algorithms

[jkrick]

Things included in this PR

1. adds pyts as a source of algorithms for light curve classification
2. cleans up the data prep section by starting to turn things into functions. I have left the functions in the main notebook instead of moving them out so it is easier to continue working with them.
3. renames the notebook to change 'lc' to 'light_curve'
4. adds functionality on how to best deal with missing data. Now there are two options. I'd like to keep both options there as I continue to develop the code.

I don't think it makes sense to do a major review on this notebook, yet. I forsee another PR before it is ready for that, but wanted to break it up to better keep track of changes.

I would take suggestions on the current functions (ie., sigma clip_lightcurves, etc.) which take in df_lc as input, change something about it, and then return that same df_lc. I think @troyraen cautioned me about doing this, but I don't see good alternatives. I could imagine renaming the light curve that it returns, so that it takes in df_lc and returns df_lc_sigmaclip. Then df_lc_sigmaclip is the input to the next function....and this goes on for a few functions....???

FYI the sktime algorithm returns a lot of FutureWarning, I put in an issue to them about that.

[troyraen]

I've look at part of this and have two comments below. I can look at the rest as well, but wanted to submit what I have now.

[bsipocz]

I went through it with a very quick review, just looking at the code and big picture and without trying to run it. Most of the comments are take it or leave it type suggestions, or just discussion starters.

[troyraen]

I looked more, but stopped at section 2.5. I'll continue from there later, but here's my comments from this round.

[troyraen]

I finished reviewing. Just one more comment below. No blockers. Thanks!

[jkrick]

Thanks for the comments, I've made changes as necessary. Some of the harder comments I have turned into issues to work on later.