Spacetime: free analytic render worker count

psf_lookup_optimization_plan.md

@@ -84,7 +84,10 @@ chosen from the validation experiments, not inferred from a performance goal.
 For every image event, compute color and final flux exactly as in the existing
 path.  Then multiply the scalar cached PSF weights by that color and flux and
 add them to the worker's existing private HDR buffer.  Preserve the current
-private-HDR ownership, memory cap, reduction ordering, and serial fallback;
+private-HDR ownership, reduction ordering, and serial fallback.  Analytic
+backends use all OpenMP render workers; a future numerical backend must opt in
+to the private-HDR memory cap when budgeting its metric slabs and evaluator
+workspaces;
 this plan does not introduce atomics, locks, or a different parallel
 partitioning.