Fix documentation error.

[twavv]

Consider the following sequence of events for a kernel.

• Client A sends execution request (id: R1)
• Client B sends execution request (id: R2)
• R1 raises an exception
• Client A receives execute_reply with content {"status": "error", ...}
• Client B recieves execute_reply with content {"status": "aborted"}

The documentation says that B should received abort rather than aborted, but the IPython kernel returns aborted (and is probably what it actually should be).

Additionally, the language around stop_on_error is confusing (I think the documentation has it backwards but I'm not sure).

[minrk]

You're absolutely right about stop_on_error being backwards. I think the original proposal was for ignore_errors which had inverted logic and the name got updated but not the description.

Would you like to send another PR for that?