Fixed common block tracking when syncing

[arkpar]

The sync module needs to track the best queued block, it does not care so much about the best imported. Before the block queue was introduced these were effectively the same. With the block queue this small change is required.

[gnunicorn]

The change looks fine,

I am just wondering about two scenarios:

1. what is said test fails to import? E.g. because the block provided doesn't actually verify. Before we have confirmed that, the block provided my be malicious. How do we ensure we recover from that scenario?
2. • but that is more general I suppose - how do we deal with forks in the chain. as from what I can see right now, the code assumes there is only ever one block at a specific height. Is that still to be sure with Aura? Isn't it allowed for two authorities to have potentially created different blocks at the same height (because the second didn't see the previous before it authored).

Arguable neither of these two problem were introduced in this PR. However if acknowledge these problem, we should create tickets for them, now?

//cc @rphmeier

[arkpar]

Originally tie syncing code was supposed to correctly download blocks from all forks. For that it keeps track of most recent common block with every peers and downloads from that point. This appears to broken now indeed as indicated by the tests. I've pushed another commit that fixes it.

As for detecting malicious behaviour the block queue keeps track of BlockOrigin for each block and indeed bans peers that give bad blocks. If such block is detected the sync process restarts.