Decommission S3 storage for the V1 server

[ryscher]

Now that we have been running on the new system for some time, we can decommission the relatively expensive S3 storage that is used by the V1 server.

Steps:

• [x] Validate that all files are in Merritt (get checksums? or generate presigned URLs?)
• [x] Decommission the mirrored bucket dryad-assetstore-west
• [x] Migrate the content from dryad-assetstore-east to Glacier
• [x] Determine conditions or schedule that would trigger decommissioning the Glacier backup

[ryscher]

I have deleted the mirror at dryad-assetstore-west and set a rule that will move dryad-assetstore-east into the Glacier Deep Archive. This content is still accessible if needed, but requires about 12 hours to retrieve. I calculate that the Glacier Deep Archive will cost about $20/month going forward, as opposed to the approximately $700/month that the standard S3 storage was costing. Based on this low cost, I recommend we leave the content in the Deep Archive indefinitely.

Note that we will continue to be billed around $650 per month for the S3 storage that holds the production Dryad content managed by Merritt, dryad-assetstore-merritt-west. We would also be able to cut the price for this by moving it to the "intelligent tiering" storage class.

[tracykteal]

Thanks for working on this. This looks good from a preservation and cost perspective.