H-091 build validation (not scientific evidence)
=================================================

Construction identities passed:
- undeformed local sub-cell convergence = 0
- synthetic reflection covariance error = 0
- synthetic non-affine cell exposes a nonzero sub-cell phase signal
- inherited H-089-R2 tiny transverse wall-sign regression passes
- Python source compiles successfully

Functional validation profile:
- R9 and R11 only
- GR-CLOSED generation completed for all five predeclared methods
- scoring and all CSV writers completed
- no method was retuned, removed, or added after seeing validation scores
- low-R validation happened to favor the inherited H-089 global control; this was NOT used to change H-091

Targeted full-resolution smoke:
- one R25 field built GR-CLOSED
- GLOBAL_H089 and all four local-sampling paths completed
- each returned 401 finite forward ticks
- LOCAL_PATH3 telemetry recorded 401 events

The real scientific run remains R25/R33 from run_H091.bat.
R39/R45 remain disabled for this telemetry-heavy exploratory pass.
