Failure to export dataset not detected, dataset marked as exported

[landreev]

This is a followup issue to #8452. The user's/installation's issue was addressed by modifying the export code to properly handle a specific metadata field becoming multiple (#8473). But there was one underlying bug that I would still like to address: despite the failure to export, the dataset was marked as exported; and then added to the OAI set. (The fact that the partial, incomplete export was left written on disk made it even worse).

Must be a straightforward enough issue of properly intercepting exceptions.

[landreev]

Added the "feature: harvesting" label, to add it to the pile of Harvesting issues that we want to fix as part of the GREI effort. (#8574)

[landreev]

This is not a specifically harvesting issue of course; but it does affect harvesting (a record like this breaks the ListRecords output; that's how it was discovered/reported).
Thinking about it, this should be addressed in both places - the export code needs to do a better job detecting errors; and the Harvesting server code needs to be able to detect the condition of the exported metadata fragment missing, without generating invalid output.

[mreekie]

sprint:

• this one is a side effect from an fix for an installation in France. That issue was addressed, and this came up during the investigation
• straightforward. Testing is time consuming.
• small

[mreekie]

• in WIP at end of pm.sprint.2022_05_11