Fix TensorRT inference issue on NVIDIA Jetson

[lakshanthad]

We need to downgrade numpy to v1.23.5 to fix a potential issue when inferencing TensorRT exported models on NVIDIA Jetson platform. I have applied the fix globally.

nvidia@nvidia-desktop:~$ yolo detect predict model=yolov8l.engine source="0"
Ultralytics YOLOv8.1.40 🚀 Python-3.8.10 torch-2.1.0a0+41361538.nv23.06 CUDA:0 (Orin, 15503MiB)
Loading yolov8l.engine for TensorRT inference...
[04/02/2024-00:13:14] [TRT] [I] Loaded engine size: 168 MiB
[04/02/2024-00:13:15] [TRT] [I] [MemUsageChange] Init cuDNN: CPU +616, GPU +584, now: CPU 1164, GPU 5393 (MiB)
[04/02/2024-00:13:15] [TRT] [I] [MemUsageChange] TensorRT-managed allocation in engine deserialization: CPU +0, GPU +167, now: CPU 0, GPU 167 (MiB)
[04/02/2024-00:13:15] [TRT] [I] [MemUsageChange] Init cuDNN: CPU +1, GPU +0, now: CPU 997, GPU 5236 (MiB)
[04/02/2024-00:13:15] [TRT] [I] [MemUsageChange] TensorRT-managed allocation in IExecutionContext creation: CPU +0, GPU +93, now: CPU 0, GPU 260 (MiB)
/usr/lib/python3.8/dist-packages/tensorrt/__init__.py:166: FutureWarning: In the future `np.bool` will be defined as the corresponding NumPy scalar.
  bool: np.bool,
Traceback (most recent call last):
  File "/home/nvidia/.local/bin/yolo", line 8, in <module>
    sys.exit(entrypoint())
  File "/home/nvidia/ultralytics/ultralytics/cfg/__init__.py", line 582, in entrypoint
    getattr(model, mode)(**overrides)  # default args from model
  File "/home/nvidia/ultralytics/ultralytics/engine/model.py", line 444, in predict
    self.predictor.setup_model(model=self.model, verbose=is_cli)
  File "/home/nvidia/ultralytics/ultralytics/engine/predictor.py", line 297, in setup_model
    self.model = AutoBackend(
  File "/home/nvidia/.local/lib/python3.8/site-packages/torch/utils/_contextlib.py", line 115, in decorate_context
    return func(*args, **kwargs)
  File "/home/nvidia/ultralytics/ultralytics/nn/autobackend.py", line 244, in __init__
    dtype = trt.nptype(model.get_binding_dtype(i))
  File "/usr/lib/python3.8/dist-packages/tensorrt/__init__.py", line 166, in nptype
    bool: np.bool,
  File "/home/nvidia/.local/lib/python3.8/site-packages/numpy/__init__.py", line 305, in __getattr__
    raise AttributeError(__former_attrs__[attr])
AttributeError: module 'numpy' has no attribute 'bool'.
`np.bool` was a deprecated alias for the builtin `bool`. To avoid this error in existing code, use `bool` by itself. Doing this will not modify any behavior and is safe. If you specifically wanted the numpy scalar type, use `np.bool_` here.
The aliases was originally deprecated in NumPy 1.20; for more details and guidance see the original release note at:
    https://numpy.org/devdocs/release/1.20.0-notes.html#deprecations

I have read the CLA Document and I hereby sign the CLA

🛠️ PR Summary

_{Made with ❤️ by Ultralytics Actions}

🌟 Summary

Improvement of NVIDIA Jetson compatibility and streamlined installation for deep learning dependencies.

📊 Key Changes

• Updated Dockerfile for NVIDIA Jetson: Specific versions of numpy are now only enforced for Jetson devices to ensure compatibility with TensorRT.
• Refined Installation Requirements: The installation process within the Dockerfile has been streamlined by grouping additional dependencies under the ".[export]" tag.
• PyProject.toml Adjustments: Added a numpy version specification for aarch64 platforms, directly addressing a known compatibility issue with TensorRT on Jetson devices.

🎯 Purpose & Impact

• 🚀 Enhanced Compatibility: Ensures that deep learning models run smoothly on NVIDIA Jetson devices by fixing a known numpy issue.
• 🛠 Simplified Setup: Streamlines the setup process for developers, making it easier to get started with deploying models on various platforms.
• 💾 Future-Proofing: By specifying dependency versions, the PR aims to minimize future compatibility issues, enhancing the user experience for both developers and end-users deploying AI models on NVIDIA's Jetson platform.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 76.62%. Comparing base (959acf6) to head (5e4b4c7).

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #9482      +/-   ##
==========================================
- Coverage   79.70%   76.62%   -3.08%     
==========================================
  Files         120      120              
  Lines       15174    15174              
==========================================
- Hits        12094    11627     -467     
- Misses       3080     3547     +467     

Flag	Coverage Δ
Benchmarks	36.30% <ø> (-0.06%)	⬇️
GPU	38.23% <ø> (ø)
Tests	71.88% <ø> (-3.48%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[lakshanthad]

@glenn-jocher FYI

[glenn-jocher]

@lakshanthad looks good!