Fix broken pytorch version constraint

[yit-b]

When trying to install pytorch3d alongside a precise version of pytorch, the constraint on pytorch's build fails erroneously due to a missing leading asterisk in the run requirements/constraints. Builds that should be allowed are blocked.

Example repro using micromamba (also repros with bare conda, but it's so slow):

CONDA_OVERRIDE_CUDA is required since the meta.yaml places a requirement that the host has cuda installed, which is not necessary, but that's a different issue.

micromamba env create -n foobar
conda activate /home/$USER/micromamba/envs/foobar
CONDA_OVERRIDE_CUDA=12.1 \
    micromamba/bin/micromamba install \
    -c conda-forge \
    -c pytorch  \
    -c nvidia \
    -c pytorch3d \
    pytorch=2.2.0=py3.11_cuda12.1_cudnn8.9.2_0 \
    pytorch3d=0.7.6=cuda120py311*

Result:

error    libmamba Could not solve for environment specs
    The following packages are incompatible
    ├─ pytorch 2.2.0 py3.11_cuda12.1_cudnn8.9.2_0 is requested and can be installed;
    └─ pytorch3d 0.7.6 cuda120py311* is not installable because it requires
       └─ pytorch * cuda*, which conflicts with any installable versions previously reported.

There is a constraint that the build begins with the string literal cuda. However, in reality pytorch builds are prefixed with the python version, followed by the cuda version. So the constraint should be pytorch * *cuda*. This change is suggested in this PR.

[conda-forge-webservices]

Hi! This is the friendly automated conda-forge-linting service.

I just wanted to let you know that I linted all conda-recipes in your PR (recipe) and found it was in an excellent condition.

[Tobias-Fischer]

Thank you @yit-b! Would you like to become a maintainer here?

[Tobias-Fischer]

@conda-forge-admin please rerender

[weiji14]

@Tobias-Fischer, we might actually want to revert this change as it would allow users to install pytorch from the pytorch channel as mentioned at conda-forge/mmcv-recipe-feedstock#31 (comment). I think @yit-b was trying to pull in a newer version of Pytorch according to #10, and wanted to relax this constraint to allow installing from the pytorch channel, but we should actually be pushing for conda-forge to have a newer version of Pytorch 2.2 rather than allowing channel mixing.

[Tobias-Fischer]

Hmm, good point, thanks for making me aware of this, I should have looked more carefully. Are you happy to open a PR?

[weiji14]

Yep, opened at #11.