fix(sync): unify ban logic in all sync processes

[mrnaveira]

Description

• New common PeerManager struct that handles peer bans. All synchronizers reuse this manager to perform the bans.
• Each synchornizer error definition (HorizonSyncError, BlockHeaderSyncError, etc.) now encapsulates the relationship (and duration) of the bans.
• Fixed logic of peer removal on long bans
• All synchronizers now handle single peer errors the same way
• All synchronizers now handle latency the same way

Motivation and Context

All sync processes (header_sync, block_sync and horizon_sync) handle bans differently., but the recommended approach is the recent block sync refactor (#5691).

The purpose of this PR is for all syncs to behave the same and ensure:

• All errors that should be banned do result in bans
• Peers are not banned for unrelated errors (database, async, etc.)
• Long bans remove the peers from the peer sync list

How Has This Been Tested?

Tests pass

What process can a PR reviewer use to test or verify this change?

Code review

Breaking Changes

• None
• Requires data directory on base node to be deleted
• Requires hard fork
• Other - Please specify

[github-actions]

Test Results (CI)

1 196 tests   1 196 ✔️  10m 29s ⏱️
     38 suites         0 💤
       1 files           0 ❌

Results for commit 40c2099.

♻️ This comment has been updated with latest results.

[stringhandler]

utACK

[SWvheerden]

pending clippy