Reorder overwritten

[stuartmcalpine]

Move the tagging of is_overwritten=True for the previous datasets to after the copy was successful

[JoanneBogart]

The change you're making looks fine, but I noticed something else nearby which maybe should be done differently. It may belong in a separate PR.
For datasets with dataset_location != "dataregistry" valid_status is left at 0. It's true there is no valid copy of the data under the control of the data registry, but I think we should distinguish such entries from those which are truly failed - either incomplete db entry, failed copy of data, or no data present where it should be when old_location was None. I think we need 3 possible values for valid_status, e.g. another bit in status to keep track of whether valid data exists or not. It would never be set for entries with dataset_location != "dataregistry", but it wouldn't mean anything was wrong in that case either. Or, alternatively, just use the combination of status and dataset_location to figure out what the valid bit means.

Either way this could require changes elsewhere in the code. If you'd rather handle that in a separate PR I can go ahead and approve this one as it stands.

[stuartmcalpine]

It looks like the valid_status variable is something left over from before, it's never actually used.

All datasets regardless of type are giving a valid=True bit if they have a successful entry. I think that's ok, they are all "valid". They can still be distinguished using the location_type column.

I've removed the left over valid_status variable.

[JoanneBogart]

Looks good. Thanks.