Update mastering-fine-tuning-with-vlm-qwen3-vl workshop to SageMaker Python SDK v3

## Summary
Migrate the `mastering-fine-tuning-with-vlm-qwen3-vl` workshop from SageMaker Python SDK v2 to v3.

## Affected Paths
- `workshops/mastering-fine-tuning-with-vlm-qwen3-vl/sagemaker_code/`

## Migration Tasks
- [ ] Update imports from framework-specific estimators to `ModelTrainer`
- [ ] Replace `sagemaker.Session()` with `sagemaker.core.helper.session_helper.Session`
- [ ] Update `get_execution_role()` import path
- [ ] Replace `output_path` with `output_data_config=OutputDataConfig(s3_output_path=...)`
- [ ] Update `image_uris.retrieve()` import path
- [ ] Update requirements.txt with SDK v3 compatible versions
- [ ] Test training jobs end-to-end

## SDK v3 Key Changes
```python
# Old (SDK v2)
from sagemaker import get_execution_role
from sagemaker.huggingface import HuggingFace

# New (SDK v3)
from sagemaker.core.helper.session_helper import Session, get_execution_role
from sagemaker.train.model_trainer import ModelTrainer
from sagemaker.train.configs import InputData, Compute, SourceCode, OutputDataConfig
```

## Related Dependabot PRs (to be closed)
- #146

Start in workshops/mastering-fine-tuning-with-vlm-qwen3-vl/sagemaker_code/ and inventory the SDK v2 imports, estimator usage, session handling, output configuration, image URI retrieval, and requirements.txt. Use the listed SageMaker SDK v3 APIs as the migration reference. Done means the requirements are compatible and training jobs run end-to-end successfully.

Written by the indexing model from the issue text.