CLAUDE: REVERT lean sum scaling - falsified by LEAN-07 (peaks blur-dominated, not fz-dominated)

LEAN-07 (stored poses, all tiles, sum-scaled input): sqrt_l0 3.22 -> 5.58
(wider, not the predicted ~2.1), corr RMS 0.82 -> 1.005. At 16x signal-to-
fat-zero the normalize amplifies the blur-decorrelated high frequencies -
the fat zero was regularizing usefully. Root of the misdiagnosis: the width
comparison used the MB-COMPENSATED oracles (FILT150A 2.08, MBEN 2.75); the
correct no-MB control (NOMB oracle) has sqrt_l0 3.58 ~ lean's 3.22 - lean
peak shapes were normal all along, and Andrey's "there was no 16x
difference" was right where it mattered.

Consequences: (1) correlator input back to the true weighted average (the
same data as the -POSE-RT-COMPOSITE debug render - single upload again);
(2) the "fz-broadened peak content pull" roll-bias mechanism is falsified -
the +0.52/+0.33 mrad roll bias needs a new suspect (remaining engine
differences vs the NOMB oracle: neighbor consolidation min_str_neib,
min_confidence gates, peak-extraction parameters); (3) MB compensation in
lean (v2) is the main quality lever - MBEN sharpens peaks 3.6 -> 2.8 and
removes the 0.5 px wobble.

Verified: mvn -DskipTests clean package OK.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

src/main/java/com/elphel/imagej/cuas/rt/CuasPoseRT.java

@@ -319,30 +319,21 @@ public class CuasPoseRT {
 		if (img_out != null) {
 			gpuQuad.execConvertDirect(false, null, 1);
 		}
-		// 3. the CPU consolidation bridge (future clt_average_sensors kernel)
+		// 3. the CPU consolidation bridge (future clt_average_sensors kernel).
+		// The TRUE weighted average is fed to the correlator (NOT scaled to the oracle's
+		// 16-pair sum): the sum-scaling experiment (LEAN-07) made peaks WIDER (sqrt_l0
+		// 3.22 -> 5.58, corr RMS 0.82 -> 1.005) - at 16x signal-to-fat-zero the normalize
+		// amplifies the blur-decorrelated high frequencies. Peak width here is BLUR-
+		// dominated, not fz-dominated (NOMB oracle 3.58 ~ lean 3.22 at 16x different
+		// amplitude); the fat zero regularizes usefully at the average's operating point.
+		// By Claude on 07/04/2026, falsified-and-reverted per LEAN-07.
 		final float [][] fclt16 = gpuQuad.getCltData(false);
-		final int [] td_counts = new int [fclt16[0].length / CuasTD.TD_CHUNK];
-		final float [] avg_td = CuasTD.consolidateSensorsTD(fclt16, td_counts);
-		// DEBUG (img_out): render the TRUE weighted average - the input-verification image
-		// (comparable to the virtual render) - BEFORE the sum scaling below. Costs one extra
-		// H2D + imclt, debug mode only. By Claude on 07/04/2026.
-		if (img_out != null) {
+		final float [] avg_td = CuasTD.consolidateSensorsTD(fclt16, null);
 		gpuQuad.setCltData(0, avg_td, false);
+		// DEBUG (img_out): render the composite the correlator sees (the weighted average)
+		if (img_out != null) {
 			img_out[0] = com.elphel.imagej.cuas.CuasMotion.perSensorImagesFromTD(gpuQuad, false)[0];
 		}
-		// Scale the average back to the per-tile SUM of sensors - the oracle inter-corr
-		// operating point (it sums the pair products unweighted, and the FZ-normalize
-		// applies the same absolute fat zero). The 1/16 amplitude of the plain average
-		// made the effective fat zero 16x larger -> broadened peaks (sqrt_l0 3.22 vs
-		// oracle 2.08 measured on identical tiles) -> asymmetric-content centroid pull,
-		// the suspected roll-bias mechanism. Per-tile count (not fixed 16) matches the
-		// oracle exactly where sensors are missing. By Claude on 07/04/2026.
-		for (int chunk = 0; chunk < td_counts.length; chunk++) if (td_counts[chunk] > 1) {
-			final int offs = chunk * CuasTD.TD_CHUNK;
-			final float scale = td_counts[chunk];
-			for (int i = 0; i < CuasTD.TD_CHUNK; i++) avg_td[offs + i] *= scale;
-		}
-		gpuQuad.setCltData(0, avg_td, false);
 		// 4. single conj-multiply: only sensor 0 (holding the average) vs the center ref
 		final double [] col_weights = scene.isMonochrome() ?
 				(new double [] {1.0}) :