Create a tensorflow CI check

## Background 

With the removal of "python-3.6-data-science-EXASOL-6.2.0" and clean up of python-3.6-data-science-cuda-EXASOL-6.2.0 we don't have any test which automatically verifies if tensorflow still compatible with our UDF's.

## Acceptance Criteria

We need to build a test which automatically creates a flavor with tensorflow (maybe as flavor_customization).

