Documentation of epsilon parameter

[cmalzer]

I have added a paragraph to the "Parameter Selection for HDBSCAN*" chapter and a demonstration of how epsilon can be used (same example as in the paper) in the new chapter "Combining HDBSCAN* with DBSCAN", which I added to index.rst.

Please feel free to correct possible grammar mistakes and to reword sentences that don't sound right to you, since English is not my first language.

By the way, I also have a Python script similar to examples/plot_hdbscan that shows how to use the epsilon parameter and allows to reproduce the GPS clusterings shown in the tutorial as well as all the other plots used in the paper. If you think that would be helpful then I could add this script to /examples, but I would then also need to add a folder "examples/data" with text files containing the data. Some of them come frome external sources (I'm using the jain, spiral and flame toy data sets
from http://cs.joensuu.fi/sipu/datasets/) and some are my own. I'm not sure where I should cite the origin of the data, maybe in a separate text file "_DataSource"?
Also, I was wondering whether it is possible to update HDBSCAN on conda-forge such that installing hdbscan with conda recognizes the new parameter?

[coveralls]

Coverage remained the same at 87.04% when pulling ea37abf on cmalzer:master into 6c1a6d4 on scikit-learn-contrib:master.

[lmcinnes]

Thanks for this Claudia; I'm very keen to get this merged and a release out that includes this, but I am going to be travelling for the next several days, so I may be a little slow in reviewing this. Please don't take my relative slowness as a lack on enthusiasm for this work. I will be getting to this as soon as I can. Thanks again!

[cmalzer]

No problem at all, it's not urgent :)

[lmcinnes]

Thanks for this -- only minor typo changes, which you can address easily.

[cmalzer]

Thanks, I just commited all of your changes.

[lmcinnes]

Thank you! I'll merge this and see if I can get a release out soon.