need better logging and error reporting when agent fails a test

Some of the intermittently failing tests are failed when jtreg simply times them out. When this happens there is little information about why the test was hanging.

Chris Hegarty wrote:

-------
Since moving to agentvm (and possibly with other modes too) we've seen this kind of issue periodically, something bad happens in a test, the harness times out and kills one of the agents. To the best of my knowledge there is no way of determining what actually caused the test to stop making progress. You may be lucky, if the test is noisy then you may be able to get something from the jtr, otherwise there's nothing useful.

I know it may not always be possible, but I guess what I'd ideally be looking for is that the harness trigger a stacktrace dump of all threads in the agentvm before killing it. This would give us a better idea why the test got stuck.
-------

Alan Bateman also mentioned having a patch for this but that it wasn't ready.