[onnx_utils] updated pytorch due to security vulnerability (mlrun#968)

* Update onnx_utils dependencies and improve test robustness

  - Upgrade PyTorch (2.6.0 → 2.8.0) and TorchVision (0.21.0 → 0.23.0) for
    better compatibility and performance
  - Update MLRun version requirement to 1.10.0 in item.yaml
  - Bump function version to 1.4.0

  Test improvements:
  - Add environment variable validation (MLRUN_DBPATH, MLRUN_ARTIFACT_PATH)
  - Add conditional test skipping based on tf2onnx availability
  - Fix cleanup function to properly remove test artifacts (model.pt,
    model_modules_map.json, onnx_model.onnx, etc.)
  - Update deprecated artifact_path parameter to output_path
  - Add explicit project context to all MLRun function calls
  - Fix PyTorch test artifact path construction

* Add conftest fixture for test environment and update notebook to PyTorch demo

- Centralize test setup/cleanup in conftest autouse fixture
- Rewrite notebook demo from Keras to a working PyTorch MobileNetV2 example

* deleted iguazio credentials

* Remove conftest.py and inline fixtures into test_onnx_utils.py

Move onnx_project and test_environment fixtures directly into the test
file to reduce unnecessary indirection for a single test module.

Author: omermaim

functions/src/onnx_utils/function.yaml

@@ -13,7 +13,7 @@ labels:
   author: Iguazio
 maintainers: []
 marketplaceType: ''
-mlrunVersion: 1.7.2
+mlrunVersion: 1.10.0
 name: onnx_utils
 platformVersion: 3.5.0
 spec:
@@ -30,13 +30,13 @@ spec:
   - tqdm~=4.67.1
   - tensorflow~=2.19.0
   - tf_keras~=2.19.0
-  - torch~=2.6.0
-  - torchvision~=0.21.0
+  - torch~=2.8.0
+  - torchvision~=0.23.0
   - onnx~=1.17.0
   - onnxruntime~=1.19.2
   - onnxoptimizer~=0.3.13
   - onnxmltools~=1.13.0
   - tf2onnx~=1.16.1
   - plotly~=5.23
 url: ''
-version: 1.3.0
+version: 1.4.0