install cache, not symlinking nvidia cuda related packages

[sawyerzheng]

• I have searched the issue tracker and believe that this is not a duplicate.

Make sure you run commands with -v flag before pasting the output.
there is not error during installation, with command pdm add torch==2.1.0 -v

Steps to reproduce

1. pdm config install.cache on
2. pdm init
3. pdm add torch==2.1.0
4. pdm run python -c 'import torch; print(torch.version)'

pdm run python -c 'import torch; print(torch.__version__)'

Actual behavior

got error as following

Traceback (most recent call last):
  File "/home/sawyer/.local/share/pdm/venvs/test-install-cache-0hLEHEEK-test/lib/python3.11/site-packages/torch/__init__.py", line 174, in _load_global_deps
    ctypes.CDLL(lib_path, mode=ctypes.RTLD_GLOBAL)
  File "/usr/lib/python3.11/ctypes/__init__.py", line 376, in __init__
    self._handle = _dlopen(self._name, mode)
                   ^^^^^^^^^^^^^^^^^^^^^^^^^
OSError: libcufft.so.11: cannot open shared object file: No such file or directory

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "<string>", line 1, in <module>
  File "/home/sawyer/.local/share/pdm/venvs/test-install-cache-0hLEHEEK-test/lib/python3.11/site-packages/torch/__init__.py", line 234, in <module>
    _load_global_deps()
  File "/home/sawyer/.local/share/pdm/venvs/test-install-cache-0hLEHEEK-test/lib/python3.11/site-packages/torch/__init__.py", line 195, in _load_global_deps
    _preload_cuda_deps(lib_folder, lib_name)
  File "/home/sawyer/.local/share/pdm/venvs/test-install-cache-0hLEHEEK-test/lib/python3.11/site-packages/torch/__init__.py", line 160, in _preload_cuda_deps
    raise ValueError(f"{lib_name} not found in the system path {sys.path}")
ValueError: libcublas.so.*[0-9] not found in the system path ['', '/usr/lib/python311.zip', '/usr/lib/python3.11', '/usr/lib/python3.11/lib-dynload', '/home/sawyer/.local/share/pdm/venvs/test-install-cache-0hLEHEEK-test/lib/python3.11/site-packages']

Expected behavior

❯ pdm run python -c 'import torch; print(torch.__version__)'
2.1.0+cu121

Environment Information

# Paste the output of `pdm info && pdm info --env` below:

PDM version:
  2.10.3
Python Interpreter:
  /home/sawyer/.local/share/pdm/venvs/test-install-cache-0hLEHEEK-test/bin/python3.11 (3.11)
Project Root:
  /home/sawyer/programs/pdm/test-install-cache
Local Packages:

{
  "implementation_name": "cpython",
  "implementation_version": "3.11.5",
  "os_name": "posix",
  "platform_machine": "x86_64",
  "platform_release": "5.15.90.1-microsoft-standard-WSL2",
  "platform_system": "Linux",
  "platform_version": "#1 SMP Fri Jan 27 02:56:13 UTC 2023",
  "python_full_version": "3.11.5",
  "platform_python_implementation": "CPython",
  "python_version": "3.11",
  "sys_platform": "linux"
}

my analysis

pdm do no symlinked cuda related packages, as in following picture, eg: nvidia-cublas package, there just a dist-info folder, not symlinked the lib folder

[frostming]

See #1732 , this will be resolved in 2.11.0, by introducing new cache methods.