[UX, aimodel] Revive rapid-turnaround plots for debugging ai models

[trentmc]

Background / motivation

We used to have a flow in https://github.com/oceanprotocol/pdr-backend/blob/main/pdr_backend/aimodel/test/test_aimodel_factory.py

Where when SHOW_PLOT = True, it has specific plots that are shown on screen when a given test is run. (And the test will fail at the end because SHOW_PLOT is True).

This is very useful for rapid iterations in developing models, and related tools (eg variable importance). Rapid = every few seconds.

It was removed as part of streamlit.

TODOs / DoD

Revive the flow. It's ok to use pyplot (vs matplotlib). It's ok to render in the browser (vs standalone window). The key thing is that it has to be light.

[trentmc]

Fixed in PR #874