Added new scan executor metric

[ArbaazKhan1]

Closes issue #5057

Introduced a new metric to track the number of exceptions that occur within the scan executor. Allows for users to monitor exceptions thrown from the iterator stack, instead of relying on parsing logs.

[keith-turner]

Would be nice to add an IT for this. Could look at the ITs that use the class TestStatsDSink for examples of how to create an IT that reads metrics.

Looked around for an IT that exercises scan executors and did not see one, its a purely performance feature and hard to test. Can look at these docs for info on setting up scan executors. Need to do something like the following in the IT.

• Before the cluster starts set some tserver.scan.executors.<executor name>.threads=<num> props in the site config to create different named scan executors.
• Create a few tables and set them to have different scan executors by setting the table.scan.dispatcher.opts.executor=<executor name> props.
• Run a scan against a table w/ an iterator that will throw an exception. Check the metrics for errors for the executor the table was configured to use.

In a way this test will also help us verify that scan dispatching is working correctly and sending things to the expected executor on the server side.

[keith-turner]

Not a change for this PR, but its interesting how this PR is tagging a metric w/ the scan executor name. Could potentially tag all scan metrics w/ the executor name.

[dlmarion]

Be sure to run the MetricsIT and other Metrics-related ITs.

[dlmarion]

I think this is looking better. A few minor comments.