Ensure the cache URLs are blanked out under all circumstances where a dataset fails to download for more than 72 hours.

[simon-20]

Brief Description
When a dataset cannot be downloaded for more than 72 hours, the file is removed from the cache and the cache URL is blanked out.

This seems to be working for most scenarios, but when a dataset fails to download because the connection times out, the dataset is successfully (and correctly) removed from blob storage, but the cache URL fields are not blanked, with the result that the unified pipeline refresher and validator continue attempting to download the file.

Severity
Medium