Cancel region update immediately on error

[paetz]

Current update behavior is like this:
If an error occurs, do not parse anymore but wait for all network requests to be executed.

It would be more efficient to cancel the network threads on error immediately.
Possible problems:

• Can we cancel a coroutine from inside?
• Cancelling a coroutine just sends a cancel request to the threads. It might be possible that the receiving & enqueuing coroutines still work for the old region while a new region's update has already been started (in case of "Update all")