Fix load_onnx! macro to use model_name parameter

[Copilot]

The load_onnx! macro was hardcoded to always load dummy_model.onnx, ignoring the $model_name parameter passed by callers.

Change

• Updated macro to use $model_name parameter instead of hardcoded path

This affects all existing callsites:

• bin.rs: now loads bin_640.onnx
• gate.rs: now loads gate_320.onnx / gate_640.onnx
• gate_poles.rs: now loads new_gate.onnx
• slalom_yolo.rs: now loads 2025Slalom.onnx

  macro_rules! load_onnx {
      ($model_name:expr, $model_size:expr, $num_objects:expr) => {{
          use opencv::core::Vector;
  
          OnnxModel::from_bytes(
-             &Vector::from_slice(include_bytes!("models/dummy_model.onnx")),
+             &Vector::from_slice(include_bytes!($model_name)),
              $model_size,
              $num_objects,
          )
          .unwrap()
      }};
  }

---

💬 We'd love your input! Share your thoughts on Copilot coding agent in our 2 minute survey.

[1upbyte]

@copilot did you eat your own brains out this will not compile you nimrod