More permissive include/exclude column specifier interpretation.

[kskalski]

Motivation

When fetching multiple dataset including some non-default columns it might be necessary to specify --include_columns value that is only present in one of the data-sets. Currently this fails as the code requires included columns to exist for each evaluated dataset.

Solution

Make --include_columns and --exclude_columns filter their values against existing columns for each data-set. This way the configuration is more permissive (non-existing columns specified in them are ignored), but also allows above use-case, which is currently not supported.

PR Checklist

• [ yes] Added Tests
• [ yes] Added Documentation
• [ non-existing include/exclude columns are silently ignored instead of failing the command] Breaking changes

[sslivkoff]

@kskalski can you give me edit access to this branch? I made a modification that checks that every included column ends up in at least one schema, otherwise error out due to bad input

[kskalski]

@sslivkoff I think it's fine to continue / re-create PR from your branch (I don't have access to members edit on my repo).

One adjustment I would make for the check you added is also verify exclude columns, since they are not required to match specific dataset now too.

[sslivkoff]

ok great, merged as #52. thanks for the good work @kskalski