Add short checkpoints stabilization without matching digests

Now a short checkpoint which is created after catch-up cannot be stabilized because it has a not even lower bound and does not have an aggregated digest and so cannot be matched with checkpoint messages from other replicas in the protocol instance.

The next checkpoint can be stabilized. However, the threshold for stashed generations of checkpoint messages from other replicas is 2. So, if a replica lags even for one 3PC-batch at the end of the next checkpoint after the short one, there is a risk that it will gather the quorum of checkpoint messages from other replicas before it completes its own checkpoint and thus an undesired catch-up will be triggered.

To avoid such undesired triggering of catch-up, we would stabilize short checkpoints using only their upper bound for matching with checkpoint messages from other replicas and without matching digests. Such the logic of short checkpoints stabilization must be added in scope of this ticket.