Add a metric for # of exceptions per scan executor

**Is your feature request related to a problem? Please describe.**
Users are using the monitor logs in order to identify when exceptions are happening in user-defined iterator code.

**Describe the solution you'd like**
Servers should emit a metric for exceptions that are thrown from the iterator stack. The metric should be tagged with the executor name that was running it.

**Additional context**
This would probably also include hadoop related exceptions. I don't know if this could get separated out at a later date.
For now, this metric would provide the same functionality that the monitor logs view does today, enabling the deprecation of the logs view in future monitor versions.