cuprated: get_info, syncer improvements

[redsh4de]

What

Improves sync-status updates and fixes a race in get_info

Why

The previous implementation had a couple shortcomings:

1. get_info read the target height separately for busy_syncing and target_height, so an update between those reads could produce a conflicting response.
2. When an invalid block caused the blockchain manager to stop the downloader, the syncer returned to waiting instead of rechecking immediately.
3. Disconnecting the peer with the highest cumulative difficulty could leave the sync status stuck on an old target height.

monerod also performs separate sync-state reads in get_info (race prone), so we avoid it by reading target_height once at the top, then derive busy_syncing and synchronized from it.

Where

cuprated, p2p-core

How

1. Re-notify the syncer immediately after a stop signal is received
2. Read target_height once at the top, busy_syncing and synchronized become derivatives of it
3. Add a peer disconnect callback, that notifies the syncer and re-checks state if a peer disconnects