Fix zonal dask memory guards and stats filtering (#1112)

* Add sweep-performance design spec

Parallel subagent triage + ralph-loop workflow for auditing all
xrspatial modules for performance bottlenecks, OOM risk under
30TB dask workloads, and backend-specific anti-patterns.

* Add sweep-performance implementation plan

7 tasks covering command scaffold, module scoring, parallel subagent
dispatch, report merging, ralph-loop generation, and smoke tests.

* Add sweep-performance slash command

* Fix zonal dask memory guards and stats filtering (#1110)

- Improve _regions_dask memory guard: use shape*itemsize instead of
  .nbytes to avoid triggering graph inspection on large arrays, and
  provide actionable error message explaining the limitation.
- Add missing memory guard to _regions_dask_cupy using GPU free memory.
- Replace iterrows() zone filtering in _stats_dask_numpy with boolean
  .isin() indexing to avoid per-row materialization.

* Add tests for zonal memory guard and stats filtering (#1110)

- test_regions_dask_memory_guard: verifies MemoryError is raised before
  .compute() when available memory is insufficient
- test_stats_dask_zone_filter: verifies zone_ids filtering works
  correctly with the new .isin() approach

Author: brendancol