forget: redesign current retention policy strategy

[AL-CT]

While the current forget command allows us to specify several options for which snapshots to retain (e.g., 2 monthly, 1 weekly, 2 daily...), it lacks flexibility. For example, it's not possible to specify that we want 2 monthly snapshots - one from the beginning of the month, and another halfway through the month - while having 3 daily snapshots that get overwritten each day, without overwriting the "stable" monthly ones.

We should reconsider the forget retention policy to better reflect our needs.

[ClaytonNorthey92]

after some discussion with @AL-CT , we're going to go with a "static" retention policy. The largest reason for this is to avoid fat-fingering possible backup deletions.

The first iteration will be this:

12 months ago
3 months ago
1 month ago
1 week ago
12 hours ago

to achieve more "exact" backups to each of these, implement the following (in pseudo-code):

for backupTime in [now - 12 hours, now - 1 week, ... , now - 12 months]:
  block = getBlockAtTime(backupTime)
  debug.setHead(block.hash)
  backup(backupTime)
  deleteOldBackup(backupTime)

this way we have "exact" timing of backups from when the backup was run upon each backup run.

we will delete the forget command

@AL-CT could you review this?

[ClaytonNorthey92]

let's force the user to configure backup timing with a similar "plugin" pattern that health checks use.

I am not sure what this will look like exactly, but it should follow a similar pattern

dave backup --geth-snaphost-times ...