Ensure NumPy/Scipy are installed from conda-forge if using conda

[swryan]

Summary

A user was getting errors after running the script in a conda environment without first installing SciPy (see below). This was traced to the fact that SciPy was getting installed from PyPI (as a dependency of pyoptsparse) and that package seemed to be incompatible with the other packages (e.g. lapack) that are installed from conda-forge by the script.

This change will install NumPy and Scipy from conda-forge before installing pyoptsparse if they are not installed, or re-install them from conda-forge if they were previously installed from PyPI. This only happens when conda is enabled.

Tests were added to the GitHub workflow to exercise these scenarios.

Also:

• one of the tests in the matrix was updated to test with SciPy 1.16, which was released recently.

Traceback (most recent call last):
  File "/Users/hschilli/miniforge3/envs/aviary/lib/python3.13/site-packages/numpy/core/__init__.py", line 24, in <module>
    from . import multiarray
  File "/Users/hschilli/miniforge3/envs/aviary/lib/python3.13/site-packages/numpy/core/multiarray.py", line 10, in <module>
    from . import overrides
  File "/Users/hschilli/miniforge3/envs/aviary/lib/python3.13/site-packages/numpy/core/overrides.py", line 8, in <module>
    from numpy.core._multiarray_umath import (
        add_docstring,  _get_implementing_args, _ArrayFunctionDispatcher)
ImportError: dlopen(/Users/hschilli/miniforge3/envs/aviary/lib/python3.13/site-packages/numpy/core/_multiarray_umath.cpython-313-darwin.so, 0x0002): Symbol not found: _cblas_caxpy$NEWLAPACK
  Referenced from: <27F2E2CD-A30E-38A6-A06C-0F5D7A0C22DC> /Users/hschilli/miniforge3/envs/aviary/lib/python3.13/site-packages/numpy/core/_multiarray_umath.cpython-313-darwin.so
  Expected in:     <3B8CB844-AEA9-37CA-901A-0BDC04DC268D> /System/Library/Frameworks/Accelerate.framework/Versions/A/Accelerate

Related Issues

• Resolves #

Backwards incompatibilities

None

New Dependencies

None