Bugfix in the ingest stage for pydantic errors on document source url.

[THOR300]

This Pull Request:

• Integrates a bug fix for catching pydantic validation errors in the ingest stage.
• The function expects a ParserInput object in the response even on failure and thus constructing this without a document source url if validation fails.
• Couldn't just move the try and except because of this.
• I don't think we need to catch validation errors on the instantiation of the ParserInput object if the document source url passes as everything else is a direct type copy. E.g. import id [str] -> document import id [str]

[diversemix]

Very nice 🍠

[olaughter]

Looks good! Only thoughts are minor, around making it easier for ourselves when we come back to this. One comment on that below, and also wondering if a more descriptive PR title would make it easier for future us?