feat: add --fields-include to unstructured-ingest

[natygyoon]

Currently, unstructured-ingest writes all top-level fields in an element with the exception of “coordinates.” This is a reasonable default. Update the command to accept a “--fields-include” in main.py so that the default is “element_id,text,type,metadata”.

[ryannikolaidis]

it's a little hard to review with the changes from the --metadata_include/exclude branch mixed in...but it doesn't look like there are any tests added to cover the new fields_include functionality, right? worth some coverage?

[ryannikolaidis]

Dropped a note on iteration and should probably add tests. Also, there are changes from the metadata_include/exclude PR mixed in here, so I can't approve that's fixed. Could either wait until that merges and rebase or make these changes against main.

[cragwolfe]

Tested with

https://github.com/Unstructured-IO/unstructured/blob/main/examples/ingest/wikipedia/ingest.sh

adding

--fields-include text,type \

Confirmed that elements in json outputs only had text and type.

Also confirmed that the outputs are equivalent to prior behavior if --fields-include is not specified.

[cragwolfe]

Please bump version and update CHANGELOG to reflect the change happening under 0.5.7-devX

[cragwolfe]

LGTM!

[ryannikolaidis]

lgtm!