Fix subsumption

[julienhenry]

I noticed a number of bugs when translating recursive subsumptive rules.

• the @Reject clauses do not need to iterate on all versions of the rule: they must always use the concrete relations from the body.
• the @delete clauses were not produced properly if the body of the subsumptive rules had atoms in the same scc. Some of the rules had multiple @delta relations, some other had none...

[codecov]

Codecov Report

Merging #2407 (d13fbc8) into master (2914f7b) will increase coverage by 0.02%.
The diff coverage is 63.63%.

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #2407      +/-   ##
==========================================
+ Coverage   77.75%   77.77%   +0.02%     
==========================================
  Files         474      474              
  Lines       31164    31164              
==========================================
+ Hits        24232    24239       +7     
+ Misses       6932     6925       -7     

Impacted Files	Coverage Δ
src/include/souffle/utility/SubProcess.h	55.00% <0.00%> (ø)
src/MainDriver.cpp	70.39% <50.00%> (ø)
src/ast2ram/seminaive/UnitTranslator.cpp	94.40% <100.00%> (-0.02%)	⬇️
src/ast2ram/utility/Utils.cpp	77.61% <100.00%> (+0.33%)	⬆️

... and 3 files with indirect coverage changes

[quentin]

Hi Julien,
I made many attempts to fix the 2 tests that crash, and I could not find the reason for the crashes yet:

• update vcpkg version
• update sqlite3 version
• synchronize sqlite3 version from vcpkg (headers) and chocolatey (dll / binaries)

It has to be something that changed in the last update of GitHub's Windows 2019 runner image.

I suggest we disable the 2 tests on Windows until we find a solution.

[julienhenry]

Yes, I don't know how to fix that... I'm going to disable SQL on Windows for the CI and see if that works.

[julienhenry]

Looks like it made things even worse...