Build pytorch test dependencies

We already have all the dependencies needed to build and run pytorch on riscv64, but we are missing some of the dependencies needed to run all the pytorch test cases. Here are a list of the missing dependencies.

onnx==1.17.0
PyNaCl==1.5.0
aiohttp==3.11.16
grpcio==1.71.0
ml_dtypes==0.5.1
pywavelets==1.7.0
pyzstd==0.16.2
scikit_image==0.22.0
scipy==1.14.1
tensorboard_data_server==0.7.2
wrapt==1.17.2
z3_solver==4.12.6.0
pyzstd-0.16.2

These dependencies either come from the .ci/docker/requirements-ci.txt file or are dependencies of the packages listed in this file. Some of the required packages are also needed by vllm and are covered in the vllm milestone. As we cannot add issues to multiple milestones and we cannot make one milestone depend on another, we'll just have to state here that the vllm milestone must be completed before we can consider this milestone to be complete.

grpcio and PyNACL are part of the Build the top 30 Python packages Milestone so is also not explicitly associated with this milestone.

PyTorch 2.7 cpu dependency tree

torch==2.7.0+cpu
├── filelock [required: Any, installed: 3.13.1]
├── typing_extensions [required: >=4.10.0, installed: 4.12.2]
├── setuptools [required: Any, installed: 70.2.0]
├── sympy [required: >=1.13.3, installed: 1.13.3]
│   └── mpmath [required: >=1.1.0,<1.4, installed: 1.3.0]
├── networkx [required: Any, installed: 3.3]
├── Jinja2 [required: Any, installed: 3.1.4]
│   └── MarkupSafe [required: >=2.0, installed: 2.1.5]
└── fsspec [required: Any, installed: 2024.6.1]

To capture this dependency tree, run the following:

python3 -m venv .venv
source .venv/bin/activate
pip install --upgrade pip
pip install torch==2.7.0 --index-url https://download.pytorch.org/whl/cpu
pip install pipdeptree
pipdeptree