Use memory_full_info() if available

On Linux, macOS and windows memory_full_info() can be used to query a process for its Unique Set Size which is a better measure for actual memory usage. Might be interesting to report this as well.

Start by locating psrecord’s existing process-memory reporting entry point and read how current memory usage is collected. Check Python’s memory_full_info() availability on Linux, macOS, and Windows, then define completion as reporting Unique Set Size alongside the existing memory measure without breaking platforms where it is unavailable.

Written by the indexing model from the issue text.