While it might appear like a trivial request to some, this is a little hairier
than what it might initially appear.

The main issue comes from the fact that runs can be resumed, so it means we
need to distinguish every execution and be able to say which subtest got
executed on which run (CI Bug Log's testsuiterun 's run ID).

This is made a little harder by the fact that a binary execution can execute
multiple tests and we need to support resuming it while still being able to say
which subtest were executed on which run.

So far, the idea is to store the current run ID in the runner's metadata.json,
then log the current run ID in the journal.

Petri, any thoughts?