[Response Ops][Reporting] Improve error logging

[ymao1]

There are several areas where we could use better error logging in the reporting process:

• Keep stack traces when logging errors - this has been useful for alerting and could provide more insight into reporting errors
• Don't repeatedly log errors. Sometimes I see Request timed out errors logged within milliseconds of each other during a single report generation. Suspect it is being logged, then rethrown and the logged again. We should make sure we're not repeatedly logging the same error as it bubbles up.
• Log more information about report doc creation. I see some generic Report failed to be created errors that provide no additional information about why it failed to be created. Stack traces and trying to pull out the error message would be helpful for debugging here.

[elasticmachine]

Pinging @elastic/response-ops (Team:ResponseOps)