mule package distributions (especially conda)

## Overview

Hi all,

I would like to ask whether it would be possible to provide official distributions of `mule`, at least on Anaconda.org (conda package), but possibly also on PyPI (pip package).

Having official distributions would significantly improve installation, reproducibility, and compatibility with other Python packages, especially in managed or HPC environments.

## Motivation

At the moment, installing mule in custom Python environments can be challenging because:

1. There are no official conda or PyPI distributions.
2. Dependencies are pinned to exact versions in `pyproject.toml`, which increases the likelihood of dependency conflicts when integrating `mule` with other packages (see related issue #18)

Providing official distributions would:

- Simplify installation for users
- Improve interoperability with the broader Python ecosystem
- Reduce the need for custom packaging workarounds
- Enable more consistent dependency resolution across platforms

## Dependencies and related work
I understand that this would likely require:

- Providing distributions for some of `mule`'s dependencies not included in this repository (e.g., `shumlib`)
- Relaxing the strict version pinning in `mule`’s dependencies (see related issue #18), since exact pins can make environment resolution almost impossible

Currently, there are [custom conda recipes](https://github.com/ACCESS-NRI/ukmo-conda-packages/blob/main/packages/mule/recipe/meta.yaml) that take `mule` from its source repository (still using SVN, with plans to move to GitHub) and distribute `mule` and its related libraries (e.g., `um-packing`, `um-spiral-search`, etc.) as conda packages. However, these are externally maintained and not officially tied to the `mule` repository.

It would be much cleaner and more sustainable if:

- The packaging recipes were maintained alongside the source code, and

- Official distributions were published directly from this project

## Questions

- Would the maintainers be open to supporting official conda and/or PyPI distributions?
- If so, would you welcome partners contributions to help set up and maintain the packaging infrastructure?
- Are there technical or policy constraints that would prevent publishing distributions?

I (along with ACCESS-NRI) would be happy to collaborate with recipe preparation, testing, and packaging if that would be useful.

Thanks for your consideration!

Davide

Start by reviewing pyproject.toml, the dependency pins, and the linked external conda recipe for mule and its related libraries. Determine whether maintainers want official conda and/or PyPI publishing, including how dependencies such as shumlib would be distributed; done requires an agreed, maintainable packaging and publishing workflow.

Written by the indexing model from the issue text.