Fix skipping of new repeated rules

[kwasielewski]

This PR add an extra check to ensure that the rule is skipped only if both lhs and rhs together are already present in the grammar.

Minimal example of a grammar affected by the change:

g = @cfgrammar begin
           S = 1 + A
           S = 2 * B
           A = 1
           B = 1
           B = 2
       end

Previously rule B = 1 would be skipped.

[sebdumancic]

thank you for the fix! this makes sense; it is a case we have overlooked.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 45.95%. Comparing base (ce15a3a) to head (1330d9c).
Report is 2 commits behind head on master.

Additional details and impacted files

@@           Coverage Diff           @@
##           master      #86   +/-   ##
=======================================
  Coverage   45.95%   45.95%           
=======================================
  Files           8        8           
  Lines         457      457           
=======================================
  Hits          210      210           
  Misses        247      247           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[ReubenJ]

Makes sense to me, too, @kwasielewski. Thanks for the contribution! Before we merge, can you add a test for this?

Also, I would like to get @THinnerichs' input, as I think we discussed this change at some point and held off on it for some reason. Now I can't remember why.

[kwasielewski]

I added a testset for this change

[THinnerichs]

For future reference:
We pushed that off to the "grammar designer" to solve. As the number of search nodes is exponential in the grammar width, we want to keep the grammar as slim as possible. If you have repetitions like that, you can always re-design the grammar such that only one of the rules is needed.

With our typed system that should not matter. We should indeed only check on entirely repeated rules, checking both left- and right-hand side.