JDAT-1228: Implement Astrowidgets API in Imviz Helper

[pllim]

In order to try and enable astrowidgets for Imviz in the future, try and see if we can get Imviz to at least support the astrowidgets API.

https://jira.stsci.edu/browse/JDAT-1228

[pllim]

Summary from meeting with @eteq on 2021-05-21:

• We will ducktype existing astrowidgets API without introducing actual astrowidgets dependency. This API list will come from what is in astrowidgets/core.py, not from https://github.com/eteq/nb-astroimage-api (also see Leveraging and synergize implementation with original design astropy/astrowidgets#136).
• Existing astrowidgets API will be divided into two groups:
  • Must-have: Must be implemented to statisfy MVP.
  • Nice-to-have: Only implement if it is trivial. Otherwise, raise NotImplementedError.
    • This includes parser APIs. Currently in Imviz, load_data delegates to the underlying parsers.py, but there are explicit methods for each data type, e.g., load_fits, load_nddata, etc. @eteq needs to think about how to consolidate these two different designs.
• @eteq wants marker APIs to be MVP but this is up in the air due to the following:
  • Can we add support for Glue's ScatterLayerState here without much work, @astrofrog ?
  • Or if we keep using the existing subset code, is that scalable to "thousands"?
  • You can argue that marker API is part of JDAT-1400 (JDAT-1400: imexamine-style photometry #531), which is not part of MVP.
  • Is this a separate ticket altogether?
• I will open a draft PR outlining the above decisions for further discussions.

[pllim]

Broken down into smaller tickets.