Fallback to remote data files while JSOC is unavailable

[nabobalis]

The goal here is to remove all of the parsing code and logic from the methods/functions that require a table either from the JSOC or offline files.

The angle here is to make the user pick the table and pass that as input as a new required input.

This avoids us having code for all the various inputs which are possible. Now the user will have to select the table they want via util functions.

TODO:

• Changelog
• Migration guide?

[wtbarnes]

I've not left very many useful comments here, but my high-level suggestion would be to refactor this such that get_response_table is unchanged and to instead just pass in the appropriate paths (using the appropriate version number) in the functions that call get_response_table. The same goes for get_error_table.

This is making me realize that all of the logic around how this information is pulled into the higher-level functions is overly complex which is largely my fault. However, this refactor should be independent of what is going on with the JSOC currently.

[wtbarnes]

I have reservations about this major of a breaking change, but I don't know of another way. My biggest issue at present is the API for getting different versions of the correction and error tables. I've added comments for a possible alternative.