Review IndexFetcher class relevance

The ``IndexFetcher`` is historically the user facade for everything related to Argo index of profiles.

But with the development of the lower level API ``ArgoIndex``, I wonder what's the overall purpose or added value of the ``IndexFetcher`` ?

This could even get confusing

We need to review its design against that of ``ArgoIndex``

- Is there any need for a high level index API ?
- Is this somehow contradictory (high-level vs index) ?
- Should we keep the erddap index fetcher ? (so slow !)
- Should manipulation of index be a sub-component of the DataFetcher ?
- In this later scenario, index is supposedly used to explore properties of a selection sampling (how many profiles, which BGC variables available, temporal/seasonal sampling ?, etc ...) before triggering data download.
- [TBC]