Add documentation page on memory and performance profiling

Users often have problems understanding where performance is slow and where memory usage is too high, especially when running out of memory in dask and running operations distributed.

We should have a documentation page explaining how to profile code for performance and memory use and track down where things are slow or memory intensive.

Start by reviewing the existing documentation structure and the profiling context described in the issue. Identify where a page on performance and memory profiling belongs, then document how users can investigate slow or memory-intensive distributed operations in Dask. Done means the page gives users a practical path for tracking down performance and memory problems.

Written by the indexing model from the issue text.