Replace per-file replication with periodic cron job to copy entire bucket at once

[tw4l]

Fixes #3342

Changes

• Remove per-file replication jobs (keeps model and some code paths required to get out and serialize the old jobs from the database, but they can no longer be run or retried)
• Add a replicate-files-cron cron job that runs on a configurable schedule (current default: every 2 hours) if any replica storage locations are configured on the instance, and launches a new copy-bucket background job for each configured replica location
• Add a copy-bucket background job that runs rclone copy on the entire primary storage bucket to a replica storage location. This will sync the replica storage with all files currently in primary without the possibility of deleting anything from the replica destination as rclone sync would, as replica deletions will still be per-file and subject to the replica deletion delay
• Stop tracking replicas on each file in the database
• Add a migration to remove per-file replica information from the database
• Remove "backed up" icons from the frontend

To note

A few nightly tests are failing (see most recent run), I believe possibly because the copy bucket job possibly doesn't have sufficient resources to start in our CI runners. We'll want to fix or change the tests before merging, but I could use another pair of eyes on the problem.

Testing

New instance

1. Set up a local Browsertrix instance with one or more default replica storage locations configured
2. Run some crawls and browser profiles
3. Verify that the cron job runs on the configured schedule and kicks off one copy bucket job per replica location
4. Verify those jobs are successful
5. Verify all files are replicated to each default replica storage location

Existing instance

1. Bootstrap an existing Browsertrix instance that already has at least one replica location configured and has previously had per-file replication jobs run
2. Verify that the new cron replication and copy bucket jobs run and are successful
3. Verify that the switch in replication jobs have not modified any filepaths unexpected in replica locations

[tw4l]

Suggested change

	successfulJobsHistoryLimit: 1
	successfulJobsHistoryLimit: 0

To reset before merging, but useful for testing

[tw4l]

Suggested change

	ttlSecondsAfterFinished: 300
	ttlSecondsAfterFinished: 0

To set to 0 before merging, but higher value useful for testing

[tw4l]

We may want to drop the commit that introduced this setting, as it doesn't seem to be making a difference for the nightly tests