Snuba-sdk

---------

- Update the Legacy converter to produce Queries that will emulate the original query as closely as possible. This way we can track migration success by ensuring the SDK and legacy calls are both producing the same Clickhouse SQL.

---------

- Added documentation for the SDK, generated from Sphinx. The docs are located at `<https://getsentry.github.io/snuba-sdk/>`_.
- The SDK is now feature compatible with the legacy JSON Snuba API, so anything that can be done with the JSON can be done using this SDK.

----------

- Some small bug fixes uncovered after doing an integration test with Sentry and Snuba.

----------

- This package was originally developed for Python 3.8+, however Sentry (the main user) is still using Python 3.6.
- The tox tests being used in Github Actions were not executing correctly, and so didn't report that this package was incompatible with 3.6.
- The package was refactored to work with Python 3.6, and the tox tests were removed from Github Actions (but left in for easier local testing).

----------

- Add a query visitor for printing, validating and translating a Query object to a SnQL query
- Allow arrays and tuples of scalars in Expressions
- Add a function that translates from JSON snuba to SnQL SDK

----------

- It is now possible to create a functioning Query, with basic validation. Also CI and release tools have all been set up.