Add additional simple support nogoods

[rtaupe]

If there are multiple ground rules in the input program (i.e., before grounding!), each with a unique ground head, we can generate completion nogoods for each of them. So far this did not happen if the unique ground heads were of the same predicate.

[codecov]

Codecov Report

Merging #226 (68c62d2) into master (1fbbbf0) will decrease coverage by 0.09%.
The diff coverage is 100.00%.

@@             Coverage Diff              @@
##             master     #226      +/-   ##
============================================
- Coverage     77.46%   77.36%   -0.10%     
+ Complexity     2312     2311       -1     
============================================
  Files           178      178              
  Lines          7782     7792      +10     
  Branches       1352     1355       +3     
============================================
  Hits           6028     6028              
- Misses         1334     1337       +3     
- Partials        420      427       +7     

Impacted Files	Coverage Δ	Complexity Δ
...t/ac/tuwien/kr/alpha/grounder/NoGoodGenerator.java	98.36% <ø> (ø)	23.00 <0.00> (ø)
.../at/ac/tuwien/kr/alpha/grounder/NaiveGrounder.java	86.66% <100.00%> (+0.54%)	75.00 <7.00> (+6.00)
...ien/kr/alpha/solver/heuristics/BerkMinLiteral.java	88.23% <0.00%> (-11.77%)	8.00% <0.00%> (-1.00%)
...a/solver/learning/GroundConflictNoGoodLearner.java	76.79% <0.00%> (-2.21%)	42.00% <0.00%> (-3.00%)
...uwien/kr/alpha/solver/NoGoodStoreAlphaRoaming.java	80.24% <0.00%> (-0.81%)	120.00% <0.00%> (-3.00%)

---

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 1fbbbf0...68c62d2. Read the comment docs.

[rtaupe]

Build fails because of #239, but I do not think the cause for this issue comes from this pull request.

[AntoniusW]

Thank you for the reminder. This PR adds certain completion nogoods, which is a good step in the right direction. There is, however, also a more principled approach in the direction of adding completion noogods currently under development (see PR #283 ). What is added here, also gets generated in #283 and more. I am currently working on the completion branch/PR (it needs a complete restructuring itself to cover more cases), with the goal of getting it ready in the next month. It is quite likely that this contribution here will be overwritten/redone/removed with #283. Do you still want it merged at this time?

[rtaupe]

Yes please, because having ready features merged to master reduces maintenance efforts when we need them in other branches.

Thanks for your work on #283 though!

[rtaupe]

I just found out that this pull request is buggy, because it does not check for multiple ground rules whose heads are of the same predicate if the heads are actually different.

For example, consider the following program:

h :- not a.
h :- not b.

With this pull request, completion nogoods for both rules are created individually, which is wrong.

Therefore, @AntoniusW I´m retracting this pull request and hope to be able to use your feature soon! I suggest you incorporate a unit test for this very condition in your branch.