Introduce RIR visitor to reduce code duplication

There are places in the code, most prominently the QIR code generation paths, where there is a mix of processing and traversal logic. To split those and avoid some code duplication, we should introduce a visitor pattern like what we use for other IRs inside our repo.

Start by reading the existing visitor implementations for other IRs in the repository, then inspect the QIR code generation paths where traversal and processing are mixed. Define the visitor boundary and confirm that the affected paths use it without duplicated traversal logic; the issue does not name specific files or tests.

Written by the indexing model from the issue text.