Testgroup verdicts

[thorehusfeldt]

Properly determine the verdicts for each testgroup.

• [x] Refactoring: introduce verdict.py
• [x] move util.VerdictTable to verdict.py
• [x] In verdict.py, provide method for Run.run to register verdicts per testcase as those are computed, inferring testgroup verdicts as these become determined. This registry overlaps with the current internal functionality of VerdictTable and needs to be unified. This is probably just class VerdictForTestItem(Dict), mapping str (testcase and testgroup names) to verdicts.Verdict.
• [x] Rewrite Run.run to leave the inference of the final (root) verdict to verdict.py
• [x] Possibly provide tree-like overview of verdicts per testcase in UI
• [ ] Possibly move Submission._get_expected_verdicts and related code to verdict.py

[thorehusfeldt]

@mpsijm : thanks for 28f9c457ef27aabe446db00670f0205adc59806e, I was writing almost the same code.

The directory mixed remains a problem and currently breaks pytest on hello.

[mpsijm]

@thorehusfeldt No problem! :smile: And it looks like the mixed directory is processed just fine, there was just an empty @EXPECTED_RESULTS@: string in one of the submissions. :joy: Let's see if that fixes it :slightly_smiling_face: (EDIT: it does! :partying_face: )

[mpsijm]

I've re-implemented lazy judging and changed it from "stop as soon as parent verdict is known" to "stop when one of the runs gets aborted" (unless -v or --table is given, of course). While I was at it, I also changed the displayed run time to be the slowest rejected test case (or, if the submission is AC, the slowest AC test case). Does anybody wants to do a double-check before we merge? :slightly_smiling_face: