The history of updates to this test got messed up so just squashing
into one commit.
Some possible regressions have been accepted, but the query is strangely
opinionated so it's just hard to say what it ought to flag.
This adds Alert annotations for alerts that seem intentional by the test
but has not been annotated with 'NOT OK', or the comment was in the wrong
place.
In a few cases I included 'Source' expectations to make it easier to see
what happened. Other 'Source' expectations will be added in bulk a later
commit.
