Refactor sampling functions

[ismael-mendoza]

It occurred to me that given how we already standardized the catalog information in catalog.py, we really don't need another layer of complexity in the sampling functions. I believe that only two sampling 'modes' are required to cover all possible cases:

• 'random': Create random blends based on the catalog of size (1, max_sources)
• 'group': Use the group_id column in the catalog to create blends. We can take the average ra, dec and use these as the center of the postage stamp.

If we really wanted to we can provide utility functions to allow the user to make cuts in their catalog (but I think users should know their catalog and can make cuts if they want to). I don't think 'global cuts' should be the job of the sampling functions.

In terms of adding a group_id column, I do think it's worth providing utility functions for doing that based on the most common use cases. Along the lines of #16

If this works out the draw_blend_generator will no longer take in a sampling function, just a flag with value either random or group. Further simplifying the UI

[ismael-mendoza]

actually like the structure in the newly refactored version. Adding something with group_id sounds like a new feature at this point