Merge nucleic/sleek-ember-seal-uady into dev

README.md

@@ -239,7 +239,10 @@ and removes each package immediately after prediction; this avoids Core ML Tools
 one full weight copy per calibration step until process exit. It prints progress while it
 runs. The candidate uses per-tensor asymmetric uint8 activations,
 per-channel symmetric int8 linear weights, and per-tensor asymmetric uint8 embedding
-weights. It fails the command if the resulting package exceeds 25 MiB.
+weights. Activation quantization is limited to floating-point linear operations; applying
+Core ML Tools' global policy also selects integer embedding-index additions and produces
+an invalid quantize operation. It fails the command if the resulting package exceeds
+25 MiB.
 
 Run the frozen gate with CPU+Neural Engine placement and compare labels directly with the
 accepted int8 ONNX artifact. Gated Core ML evaluation fails closed without