tests: exercise the fused decode in every cossim harness

Each harness re-runs the C++ side with --codec-fused after the
buffered run and gates on codes-full.bin equality between the two:
the decode path cannot change the predictor, so any divergence is a
bug. The fused run's perf lines land in the log as [Perf Fused] next
to the buffered ones, so every grid cell carries the buffered vs
fused comparison per backend and quant.

The fused audio itself is not scored here: greedy amplitudes vanish
in the streamed PCM_16 and the fused FP envelope has its own
validation against the stream path.

tests/debug-customvoice-cossim.py

@@ -181,5 +181,7 @@ def main():
     n = min(aa.size, ab.size)
     print(f"[Cossim] WAV stft_cos: {cc.stft_cos(aa.ravel()[:n], ab.ravel()[:n]):.6f} samples: {n}")
 
+    cc.fused_pass(cmd, text, DUMP_CPP)
+
 if __name__ == "__main__":
     main()