What change would you like to see?

The replay.json for crawls and collections should be optimized to look up and presign bucket paths in bulk.

Context

Loading a crawl or collection with large number of WACZ files can be slow initially, since each file's path is looked up in the presign collection, and then if needed, each file is signed individually. This results in collections with hundereds of WACZ files to be slow on first load.

This can definitely be improved with batching querying and batch presigning.