Maintenance mode

We need to be able to switch Mistral cluster (in fact, engines) to something like "maintenance mode" so that while in this mode Mistral would not allow to new workflows to start.

The motivation behind it is one the following:
* If have a working cluster and we decide to reconfigure Mistral (ports, DB etc.) then we need to wait till all currently running workflows are completed and then we could safely shut down the cluster and restart it
* If there was a failure in the cluster (e.g. part of executors died in the middle of processing actions) and part of running workflows/tasks went to ERROR state, we should be able to switch to maintenance mode and manually go over failed workflows/tasks and apply resolution (clear status, rerun etc.)

This work is also related with UI because in UI we'll need to implement needed controls to resolve problems in existing running (if we know they will never finish) or failed workflows/tasks.