Migration to bioconda

[sjanssen2]

Improvement Description
I finally was able to clean up Siavash's source code and created a bioconda recipe for SEPP, producing the packages at https://anaconda.org/bioconda/sepp

Note that this package does NOT contain the default Greengenes 13.8 99% reference (which consists of three files a) alignment b) tree c) info file.) In the future, we also want to support alternative references like SILVA.

Proposed Behavior
I wonder how we best do this? I see the following options:

1. create conda reference packages for GG and / or SILVA
   pro: no changes to current behaviour of qiime2
   con: where to host? would bioconda accept that?
2. provide as qiime2 Data resources
   pro: smaller downloads when qiime2 gets installed, easy to host
   con: user need to do extra work when a) install and b) execute since file paths for all three files need to be provided

Questions
Any thoughts @thermokarst @antgonza ?

References

1. recipe
2. https://anaconda.org/bioconda/sepp

[antgonza]

Just wondering, what will be your thoughts about adding a command/method to download the available references and set them in your environment? Something like: qiime fragment-insertion --show-available-references; qiime fragment-insertion --download-reference XXX; qiime fragment-insertion --show-local-references-locations ...

[thermokarst]

I think I am leaning towards option #2. @ebolyen and I recently discussed some potential machinery for fetching and caching data assets on a per-deployment basis, which might ease the user-burden (so, something similar to what @antgonza is proposing).