Allow to update or replace failed resources custom

Recently I developed some custom resource plugins to integrate our production, and the mechanism of replacing failed resource is so bad, the state of resource is UPDATE(*)_FAILED in heat, but actually it is in active underlying, what's more is that we can't to re-create some resources with same properties as old ones, because there is nothing we can do before replacing, e.g we do something for nova server, detaching port from the old server to make sure we can create a new server and attach the same port. But not every resource can apply this mechanism.

I have proposed a bug https://bugs.launchpad.net/heat/+bug/1490134, but it's better to register a blueprint to trace, also I will propose the series for other resources.