Investigate effort of testing Python

[wg102]

This relates to #831

The idea is for CWA to be implemented, we may need to recode it like we did https://github.com/StatCan/jupyter-apis and have the backend in Go instead of Python.

From what was explained, this was due to a performance issue when using a large amount of namespaces (400+).

The idea of the ticket is to investigate how much of an effort it would be to test the python backend in our dev environment in a way that simulates prod enough to be viable results. In other words, with traffic/namespaces/IO similar to prod.
Versus the effort it would be to 'simply' recode the backend and new features in go.

The concern is really to have dev accurate enough for viable efforts.

[rohank07]

After a brief discussion with Will, verified we would need to keep the backend in Go. It will still not address the performance issue in the implementation of kubernetes-client.