The De Bruijn index system's hardest case is a function with many let @Int = ... bindings where indices shift with each binding. The current conformance suite (ch03_slot_indexing.vera) covers basic cases but not deep chains.

Write conformance tests with 5+ sequential Int let-bindings and verify the indices resolve correctly at each depth. This is the scenario most likely to confuse both LLMs generating code and the compiler when adding new let-binding logic.

Also test the interaction with match arm bindings, where indices are further shadowed.