[CIR] Fix ptrtoaddr vs ptrtoint lowering behavior after #210729

#210729 changed the behavior of how we're lowering pointer behavior in 'classic' codegen.

This change in behavior was NOT captured by CIR. #216963 updated the tests to match it.

Whatever change happens here has to analyze what happened/what logic applied to the original patch, and mirror that in CIR.

Start by reading #210729 to understand the classic codegen lowering change, then compare it with the CIR behavior described here. Use #216963 to identify the tests that were updated, and consider the work complete when CIR mirrors the original lowering logic and those tests reflect the matching behavior.

Written by the indexing model from the issue text.