Ignore eml files when processing extracted reports

[moorereason]

Given input.delete = no

After eml files are processed and attachments have been extracted, the convert/parse phase attempts to reprocess the eml files as reports and logs an error:

[ERROR] files: mimetype text/plain; charset=utf-8 not supported in file dmarc_files/foo.eml, skip

We should ignore the eml files in this phase.

I plan to submit a PR shortly to resolve this issue.