dcp4 probe: add --cp-kv-cache-interleave-size 4 (GLM C4 DCP requirement)

recipes/glm53-flash-nvfp4-r17abi-4x-dcp4/recipe.toml

@@ -104,6 +104,7 @@ args = [
     # 92.72 GiB weights/node; pinning also keeps warmup off the swap path
     "--kv-cache-memory-bytes", "10737418240",
     "--decode-context-parallel-size", "4",
+    "--cp-kv-cache-interleave-size", "4",
     # No --kv-cache-dtype: let the r12 fork pick its default (the explicit
     # fp8 / fp8_ds_mla both failed auto-selection on this build).
     # The GLM5Next model wires the B12X sparse-MLA backend (sm_120/sm_121)