Hepdata-converter-ws

What's Changed
* Changes needed to work with hepdata-converter v0.3.0 release such as supporting two YODA formats.
* Switch from Travis to GitHub Actions by alisonrclarke in https://github.com/HEPData/hepdata-converter-ws/pull/15
* Further upgrades to CI workflow and dependencies such as Flask.

**Full Changelog**: https://github.com/HEPData/hepdata-converter-ws/compare/0.2.0...0.3.0

This release fixes a long-standing bug (see 14) where an argument `r:gz` is incorrectly passed to `tarfile.open` as a pathname `name` instead of the keyword argument `mode`. This does not seem to have consequences unless the file cannot be opened, when a `FileNotFoundError` exception can be returned instead of the expected `ReadError`.

This release adds a new `get_data_size` method that gets the size in bytes of the JSON data that would be sent to the converter. This could be useful in checking that a maximum payload size imposed by a web server is not exceeded.

This is the first release for Python 3. It also adds support for Sentry.

* Adds unit tests to check that `validator_schema_version` option is passed correctly.
* Remove `--dry-run` option from `SetupTestCase` in `test_setup.py`.
* Remove circular dependence on `HEPData/hepdata-converter-ws-docker` repository.
* Run Travis CI tests inside a `hepdata-converter` (not `hepdata-converter-ws`) Docker image.
* Require `hepdata-converter>=0.1.35,<0.2` and `flask>=1.1.1,<2` in `setup.py`.
* Remove temporary files written to `/tmp` following a failed conversion.
* Prevents file system of the Docker container being filled (Jan. 2020).
* Tidy `README.md` file, add badges, include as `long_description` of `setup.py`.

This release is not strictly needed, since the only changes with respect to version 0.1.5 are in the tests, but I'm tagging it to mark the first Python 3 deployment of the hepdata-converter packages.