Adding additional metrics during instrumentation of clients (#191)

[bsieber-mozilla]

In this PR we can now push metrics of successful jira calls and exceptions

@leplatrem does this make sense? Can success rate already be determined by existing metrics?

[leplatrem]

@leplatrem does this make sense? Can success rate already be determined by existing metrics?

I asked Ben here whether having these high level operations are enough (instead of client SDK methods calls).

If yes, then we should probably only add a fail operation, that we log when an exception is raised in runner.
If not, then maybe adjust this PR to add logger calls instead of StatsD metrics, since BigQuery seems to be the way to go for this kind of reporting.