Improve inference for context sensitive functions within reverse mapped types

Improve inference for context sensitive functions within reverse mapped types #34732