While the XML coverage report is useful for consumption by other tools, such as
currently by codecov.io in CI, it's not very useful for humans reviewing the immediate
impact of changes on coverage during local development or while monitoring CI output. I
don't think running the text report takes much more time so I don't see a downside
here.
Run with "tox -e coverage". Uses a new helper
module (allmydata.test.run_trial) to let us import+execute trial without
knowing exactly where the "trial" binary lives, which helps with using
"coverage run" under tox.
