Add Docker build test workflow and update environment dependencies

[adamjtaylor]

This pull request introduces a new GitHub Actions workflow to automate Docker image build and testing on pull requests, and refactors the Docker build process to improve dependency management and compatibility with NumPy 2.x and zarr 3.x. The main changes are focused on ensuring that dependencies are installed via conda and pip in environment.yml rather than directly in the Dockerfile, and on verifying the compatibility of critical packages in the CI pipeline.

Continuous Integration Improvements

• Added a new GitHub Actions workflow .github/workflows/build-test.yml to build the Docker image and run smoke tests on dependencies and critical package imports for every pull request affecting Docker-related files. This ensures that the Docker image builds successfully and that key dependencies (including NumPy 2.x and zarr 3.x) are correctly installed and compatible.

Dependency Management Refactoring

• Moved installation of openslide-python, opencv-python-headless, synapseclient, mantel, ome_types, and minerva-lib-python from the Dockerfile to the pip section of docker/environment.yml, ensuring a more reproducible and maintainable environment specification. [1] [2]
• Updated the numpy dependency in docker/environment.yml to require version 2.x and explicitly added zarr>=3.0.0 to ensure compatibility with the latest versions.
• Removed redundant or duplicate dependencies (zarr, matplotlib) from the main dependencies list in docker/environment.yml to streamline the environment.

These changes help ensure the Docker environment is robust, up-to-date, and that future pull requests are automatically validated for dependency compatibility.