Add SLLambdaEq learning algorithm

[FredrikTaquist]

This PR adds the new SLLambdaEq learning algorithm, an extension of SLLambda optimized for the equality theory.
The new algorithm improves scalability by employing additional restrictions than those implemented in SLLambda, which optimize for equalities between between parameters in suffixes and prefixes when making tree queries.
The SLLambdaEq algorithm is implemented on top of SLLambda, with new functionality added to the ClassificationTree data structure in order to make use of the new types of suffix restrictions.

In addition, this PR fixes a number of bugs in SLLambda that were encountered during the implementation of SLLambdaEq, including (but not limited to) the following:

• Fixes a bug where the wrong data values would be compared during counterexample analysis.
• Fixes a bug in which SDTs were incorrectly deemed inequivalent in cases where there was an equality between data values in one prefix but not the other.
• Fixes a bug when instantiating a data symbol from a guard where relations with prior parameters of the symbol were not correctly instantiated.
• Fixes a bug involving restrictions for revealing missing registers when resolving Register Closedness, where the wrong set of registers were used.