FIX: Rule conversion for EFDT and VFDT trees

[sankhya10]

Changes are linked to:

• Support decision trees that have a single root node with no children #7
• Investigate why EFDT trees with different number of nodes have a similarity score of 100% #8
• Support VFDT trees #9

Done:

• The issue with a decision tree that has a single root node with no children is forced to have an antecedent in order to calculate similarity. For the sake of the discussion, two single root nodes (with arbitrary labels) will have a 100% similarity in reference to the paper.
• For example originally for the root node the label would be $\text { r0: } \rightarrow 1$. Now the rule would changed
  to $\text { r0: attrname }<=0 \rightarrow 1$
• The rules for EFDT and VFDT trees are now based on the condition of the tree and based on the rules. Originally it was $<=$ for all rules.
• Similarity metric can now compute similarity between an efdt tree and a vfdt tree.