Fix RWS with score mode MAX_SUBTASK

[lucach]

Tasks with a score type that was not group (i.e., without subtasks)
were not handled correctly.

We consider them as having one subtask (as already done elsewhere in
CMS).

Important note
I think that the proposed change can work in the event of an unscored submission (which should have 0 as score), but please double-check what is being done inside those lines. Feel free to modify them as needed.
They were disputed since their introduction and have led to a number of subtle bugs in these last months. 😂

---

This change is 

[wil93]

Mmm I feel like a better solution would be to avoid altogether having tasks with score_mode = 'max_subtask' and no subtasks.

A DB constraint maybe?

[codecov]

Codecov Report

Merging #1111 into master will increase coverage by 0.11%.
The diff coverage is 0%.

@@            Coverage Diff             @@
##           master    #1111      +/-   ##
==========================================
+ Coverage   62.02%   62.14%   +0.11%     
==========================================
  Files         230      230              
  Lines       16588    16588              
==========================================
+ Hits        10289    10308      +19     
+ Misses       6299     6280      -19

Flag	Coverage Δ
#functionaltests	45.97% <0%> (+0.2%)	⬆️
#unittests	43.27% <0%> (ø)	⬆️

Impacted Files	Coverage Δ
cmsranking/Scoring.py	57.84% <0%> (ø)	⬆️
cms/db/filecacher.py	78.36% <0%> (-0.99%)	⬇️
cms/service/ResourceService.py	57.99% <0%> (-0.92%)	⬇️
cms/db/fsobject.py	80.46% <0%> (-0.79%)	⬇️
cms/io/service.py	68.9% <0%> (-0.61%)	⬇️
cms/grading/Sandbox.py	68.23% <0%> (-0.37%)	⬇️
cms/grading/Job.py	89.57% <0%> (+0.47%)	⬆️
cms/db/util.py	52.59% <0%> (+0.74%)	⬆️
cms/service/EvaluationService.py	67.55% <0%> (+1.06%)	⬆️
cms/service/ProxyService.py	58.51% <0%> (+1.06%)	⬆️
... and 7 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update bfb359b...1fed420. Read the comment docs.

[lucach]

That is another option, it depends on how you watch the issue.

I chose this path because it is how CMS treats this situation elsewhere in the codebase (see https://github.com/cms-dev/cms/blob/master/cms/grading/scoring.py#L247-L249 ).

[stefano-maggiolo]

Yeah, I see. Seems to make sense, but I'll take another look for the corner case you pointed out. But why does this bother you? Are you using this combination of configurations?

[stefano-maggiolo]

Yes, I think it makes sense, since in reset_history the score is set to 0 by default (then changes are applied).

[lucach]

But why does this bother you? Are you using this combination of configurations?

Not intentionally (but it happened by mistake during the final round of the Italian Olympiad in Teams 😆 )

Thanks for merging the fix, at least now CMS behaves properly. For the future some more profound changes, as those suggested by @wil93, can be implemented.