There's currently no built-in propagation of constant evaluations in the instructions or remapping of operators if certain logical operators select one result over the other. The propagated results are only taken into consideration for branches and switches, but other instructions are never patched with the results of the evaluation.
This would help to further eliminate some instructions where the chain of operations is not deleted while a large part of the tree can evaluate as a constant.