Redbiom

PR https://github.com/biocore/redbiom/pull/122 in particular included quite a few issues, as well as performance boost for fetch, a new API method for fetching a Qiita study, clean up of warnings, removal of Py3.6 from build testing (the environments won't create) and addition of Python 3.10 into the matrix.

Specific other items addressed in https://github.com/biocore/redbiom/pull/122 include:

* Allow for resolution of ambiguities while retaining the artifact ID https://github.com/biocore/redbiom/issues/119
* A particular sample always crashes for me https://github.com/biocore/redbiom/issues/117
* It would be nice to have an easy option to pull a specific qiita study https://github.com/biocore/redbiom/issues/116
* redbiom summarize taxonomy should handle no hits gracefully https://github.com/biocore/redbiom/issues/43
* redbiom search metadata NUMBER should fail gracefully https://github.com/biocore/redbiom/issues/44
* Fixing instructions in readme https://github.com/biocore/redbiom/issues/46
* If context creation fails, error gracefully https://github.com/biocore/redbiom/issues/52
* search for phrases https://github.com/biocore/redbiom/issues/50
* Where queries should support "in" correctly when the left hand is a string https://github.com/biocore/redbiom/issues/41

What's Changed
* API/PERF: allow avoiding fetch of taxonomy by wasade in https://github.com/biocore/redbiom/pull/114
* add msgpack by wasade in https://github.com/biocore/redbiom/pull/115
* ping! by antgonza in https://github.com/biocore/redbiom/pull/118

**Full Changelog**: https://github.com/biocore/redbiom/compare/0.3.8...0.3.9

Important bulk load operation to reduce load/refresh times.

PR 113 carries an important fix where samples with a `.raw` suffix were not being loaded.

As the title says :)

This micro-release includes additions that allow for the resolution of ambiguities at data and metadata retrieval through two different means. The first is to retain the sample with the most sequence data, the second is to merge ambiguous samples together.

This release also includes a bug fix where periods were not being escaped, and in some extreme examples getting interpreted by webdis as denoting an altered return type over the default of JSON.

Finally, we now also allow for the retrieval of specific metadata columns!

This is primarily a bug fix release to allow indexing of "/" characters at cache generation. Strings containing these characters were previously omitted as they are reserved in the case of a URL. These values are now formally encoded using `urllib`.

This release also includes support for optionally hashing features on `fetch` -- thanks antgonza!