Peak memory usage for parallel tests

## Feature Request

When running the memray plugin over multiple tests in parallel (using `pytest-xdist`), it shows only the N tests that used most memory.

While that's great, I would like to see a summary of the **total** memory used by all tests together (especially I'd like to know what's the peak memory usage).

A possible solution might be using the vanilla memray, but I'm not sure how to do that, and regardless of that - I think it's useful to show this information in addition to the existing one.

Thanks!

Start by tracing how the plugin handles pytest-xdist workers and currently selects the N highest-memory tests. Define how total and peak memory should be aggregated across workers, then add coverage for a parallel run and verify the report includes both the existing top-N results and the new summary.

Written by the indexing model from the issue text.