Make harvester more defensive against bad data

[swirtSJW]

User Story

As a person monitoring dataset harvesting, it would be nice if the harvester did some additional pre-harvest validation to prevent bad data from being imported.

Acceptance Criteria

• Prevents harvesting if data is not UTF8
• Prevent harvesting if BOM character is present (or maybe silently remove it?)
  • Likely need this to be an option that is off by default as apparently no all data streams care.
    There may be some additional checks needed but I need to determine if they are generic enough to apply to all, or if they perhaps need to be optional.

[swirtSJW]

I am still sorting out what is specific to PDC and what is generic enough for broader use.

[dafeder]

At minimum, I think it makes sense to document how to implement this kind of "kill switch" in a Harvest Extract class.

[swirtSJW]

I will be working on this in the next sprint.