Background task to re-check checksums

[mtbcarroll]

After a file has been uploaded it may be corrupted through some later glitch on the server. Uploaded files should have their checksums in the hash column of the originalfile table re-checked periodically by some background job, and the user notified of any detected problems. Files without a checksum should probably have one calculated so that subsequent corruption can at least be detected.

It may make sense for the model to include originalfile columns for "timestamp of when the hash was first calculated", "timestamp of when the hash was last correct", and "if the file now appears to be corrupted".

[mtbcarroll]

(Actually, the model probably doesn't need the timestamps if the relevant information can be extracted easily from a system log.)

[mtbcarroll]

Bumping up the priority here a little given the advent of in-place import.

[jamoore]

I wonder if this and similar tasks might need to show up on the "priorities list" board sooner rather than later... Though we can certainly add the check post-haste, the larger usability issue is that we will need to provide that feedback to the appropriate person (sysadmin, groupadmin, data owner). And that work will need to be scoped, etc.

I added ​https://trello.com/c/iJvGlYnW/130-background-data-checks but prioritization still needs to happen.