Revert "wheels CI: update to torch 2.10 for 'oldest' configuration (#5485)

[jameslamb]

Reverts #5485

We can now install torch 2.9 CUDA 12.9 wheels again, the issue on the PyTorch pip index has been resolved: pytorch/pytorch#179821 (comment)

[jameslamb]

This is a wheels-only change and all the wheels CI is passing... I'm going to admin-merge this, so there's one less PR to juggle here.

The conda test failures are unrelated

_____________ ERROR collecting cugraph/pytest-based/bench_algos.py _____________
ImportError while importing test module '/__w/cugraph/cugraph/benchmarks/cugraph/pytest-based/bench_algos.py'.
Hint: make sure your test modules/packages have valid Python names.
Traceback:
/opt/conda/envs/test/lib/python3.11/importlib/__init__.py:126: in import_module
    return _bootstrap._gcd_import(name[level:], package, level)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
cugraph/pytest-based/bench_algos.py:9: in <module>
    from pylibcugraph.testing import gen_fixture_params_product
/opt/conda/envs/test/lib/python3.11/site-packages/pylibcugraph/__init__.py:15: in <module>
    import pylibcugraph.comms
/opt/conda/envs/test/lib/python3.11/site-packages/pylibcugraph/comms/__init__.py:4: in <module>
    from .comms_wrapper import init_subcomms
E   ImportError: /opt/conda/envs/test/bin/../lib/libcugraph.so: undefined symbol: cudaLibraryGetKernel, version libcudart.so.12

(build link)

That's being worked on here: #5496