Failed installation in a container

Hi, I am trying to install lsdb inside a pytorch container on NERSC perlmutter supercomputer. The link to the image is [here](https://hub.docker.com/layers/nersc/pytorch/24.08.02/images/sha256-a69f778e39cf9f9721adc8e40d3da11788cb2d3de2a9c00104b99802d53531b5). This is essentially the same as the ones hosted on nvidia official release. The container has python 3.10.

Although my installation was finished with pip error messages on a dependency issue, it failed to import lsdb and hats when I inspected it interactively. It seems that `hats` require `numpy<3, >=2`, while many modules lsdb depends on do not support numpy > 2. This seems to be a compatibility issue, could you please fix it? I attached my containerfile and the build message below.

[build-pip.out.txt](https://github.com/user-attachments/files/22183358/build-pip.out.txt)
[Containerfile-lsdb.txt](https://github.com/user-attachments/files/22183359/Containerfile-lsdb.txt)

Start with Containerfile-lsdb.txt and build-pip.out.txt to reproduce the installation in the Python 3.10 PyTorch container and identify the conflicting numpy requirements between hats and lsdb dependencies. Done means lsdb and hats install without dependency errors and both import successfully in that container.

Written by the indexing model from the issue text.