Currently a lot of the tests are integration tests with the test server, which makes them rely on internet and run veeeery slowly.
Maybe we should add more mocking to allow for local tests and run the integration tests only sometimes / on the CI?
@rquintino did some work on having a local installation of OpenML via Docker, more information is in the SLACK channel and this PDF, maybe that's also a solution.
Currently a lot of the tests are integration tests with the test server, which makes them rely on internet and run veeeery slowly. Maybe we should add more mocking to allow for local tests and run the integration tests only sometimes / on the CI?