refactor(src): standardize os.environ usage in inference upload script (#194)

- add set_env_defaults import from training.utils in main()
- centralize default values for 8 environment variables
- replace os.environ.get with os.environ[] for defaulted vars
- keep os.environ.get for optional and dependent vars

🔧 - Generated by Copilot

Author: WilliamBerryiii

src/inference/scripts/upload_artifacts.py

@@ -90,10 +90,10 @@ def upload_to_mlflow(
             )
             mlflow.log_params(
                 {
-                    "num_envs": os.environ.get("NUM_ENVS", "4"),
-                    "max_steps": os.environ.get("MAX_STEPS", "500"),
-                    "video_length": os.environ.get("VIDEO_LENGTH", "200"),
-                    "inference_format": os.environ.get("INFERENCE_FORMAT", "both"),
+                    "num_envs": os.environ["NUM_ENVS"],
+                    "max_steps": os.environ["MAX_STEPS"],
+                    "video_length": os.environ["VIDEO_LENGTH"],
+                    "inference_format": os.environ["INFERENCE_FORMAT"],
                 }
             )
 
@@ -267,14 +267,27 @@ def main() -> None:
     if src_root and src_root not in sys.path:
         sys.path.insert(0, src_root)
 
-    task = os.environ.get("TASK", "unknown")
-    export_dir = Path(os.environ.get("EXPORT_DIR", "/tmp/exported"))
+    from training.utils import set_env_defaults
+
+    set_env_defaults({
+        "TASK": "unknown",
+        "EXPORT_DIR": "/tmp/exported",
+        "ONNX_SUCCESS": "0",
+        "JIT_SUCCESS": "0",
+        "NUM_ENVS": "4",
+        "MAX_STEPS": "500",
+        "VIDEO_LENGTH": "200",
+        "INFERENCE_FORMAT": "both",
+    })
+
+    task = os.environ["TASK"]
+    export_dir = Path(os.environ["EXPORT_DIR"])
     metrics_dir = Path(os.environ.get("METRICS_DIR", str(export_dir / "metrics")))
     checkpoint_uri = os.environ.get("CHECKPOINT_URI", "")
     blob_account = os.environ.get("BLOB_STORAGE_ACCOUNT", "")
     blob_container = os.environ.get("BLOB_CONTAINER", "")
-    onnx_success = os.environ.get("ONNX_SUCCESS", "0") == "1"
-    jit_success = os.environ.get("JIT_SUCCESS", "0") == "1"
+    onnx_success = os.environ["ONNX_SUCCESS"] == "1"
+    jit_success = os.environ["JIT_SUCCESS"] == "1"
     timestamp = datetime.utcnow().strftime("%Y%m%d_%H%M%S")
 
     onnx_metrics, jit_metrics = load_metrics(metrics_dir)