[RFE] Handle Timeouts Generically

[Vladyslav Drok]

When a request is received via RPC, some work is started asynchronously by the Conductor process. At this time, there is no process watching to see how long that work has taken and potentially stop it after a configurable timeout.

Because drivers may implement their respective interfaces in different ways, we need a general way for the Conductor to ask the driver if an operation has taken "too long." The set of timeouts should be standardized across drivers. Individual timeouts should be configurable per-driver.

This also requires that drivers advertise which operations are interruptable, and a means to interrupt those that are.

Related to:
- https://bugs.launchpad.net/ironic/+bug/1526352
- https://blueprints.launchpad.net/ironic/+spec/abort-deployment

[Jim Rollenhagen (jim-rollenhagen)]

This will need a spec