fix: remove explicit port from NF3 hybrid recipes (use default)

recipes/glm52-nf3-hybrid-context/recipe.toml

@@ -41,7 +41,6 @@ engine = "vllm"
 name = "GLM-5.2 NF3-Hybrid CONTEXT LANE (vLLM)"
 docker_image = "vllm-nf3-hybrid:probe"
 model = "madeby561/GLM-5.2-MXFP8-NVFP4-NF3-Hybrid"
-port = 8210
 required_nodes = 4
 tensor_parallel_size = 4
 pipeline_parallel_size = 1

recipes/glm52-nf3-hybrid-fast/recipe.toml

@@ -41,7 +41,6 @@ engine = "vllm"
 name = "GLM-5.2 NF3-Hybrid FAST LANE (vLLM)"
 docker_image = "vllm-nf3-hybrid:probe"
 model = "madeby561/GLM-5.2-MXFP8-NVFP4-NF3-Hybrid"
-port = 8210
 required_nodes = 4
 tensor_parallel_size = 4
 pipeline_parallel_size = 1