Indexing optimization

[hectorcorrea]

We are currently indexing the file list for each record twice, one as part of pdc_describe_json_ss and another as files_ss. This is not a problem for small datasets but when we have datasets with 60K files this is rather inefficient.

There is really no need to index the second field (files_ss) since pdc_describe_json_ss has the file list already.

Making this change will require calculating the file list (rather than storing it in Solr) from the data in pdc_describe_json_ss while keeping in place the logic in the https://github.com/pulibrary/pdc_discovery/blob/main/config/traject/pdc_describe_indexing_config.rb#L244-L261

Looks like staging is sometimes having issues processing some of our large datasets. Sometimes they succeed sometimes they don't (see details in Honeybadger https://app.honeybadger.io/projects/95072/faults/116339390). The fix in this issue might help here.