This commit adds more full range caching to the STM such that we'll
be able to prefetch keys and ranges on the kvdb level. This will allow
us to reduce roundtrips to etcd do fast iteration of small ranges.
This commit builds on the ideas of @cfromknecht in lnd/5153. The
addition is that the design is now simpler and more robust by queueing
up everything, but allowing maximal parallelism where txns don't block.
Furthermore the commit makes CommitQueue.Done() private essentially
removing the need to understand the queue externally.
