Investigate the possibility of backfilling runs

[MarcusNoble]

Look into the possibility of backfilling experiment runs so that reduction jobs can be performed on past data.

Things to consider:

• Disk space
• Ascending run numbers
• Resources (run out of cycle?)

Investigation results:

• Disk space is fine for the time being as only a few instruments are going to be using it for now. Also, as the interface limits results to the past 3 years it doesn't make sense to backfill any further back than that.
• The order of run numbers isn't an issue.
• Resources are an issue and back filling a lot of run numbers could prevent live data from being run right away. Currently backfilling will be on a request basis and run out-of-cycle. Going forward priority queues could be used to help with this. See #10794