• Package for Python 3.13.
• Updated dependencies.

• Package for Python 3.10, 3.11, and 3.12.
• Updated dependencies and workflows.

• Add Krack-Köster radial grid (radial_grid_kk).
• Rename radial_grid to radial_grid_lmg.
• Make it possible to evaluate LMG grid by basis set name (radial_grid_lmg_bse).

• Better error message for invalid inputs to angular_grid.
• Improve package metadata.

• Test and build also for Python 3.9.
• Export also version string in the Python module.

The API changed for easier maintenance and simpler use:

• No initialization or deallocation necessary.
• One-step instead of two steps (since the radial grid generation time is
  negligible compared to space partitioning, it did not make sense anymore to
  separate these steps and introduce a state).
• alpha_min is given as dictionary which saves an argument and simplifies
  explaining the API.
• The library now provides Rust and Python bindings. It used to provide C and
  Fortran bindings. The C/Fortran code lives on on the `cpp-version\ branch. I might bring the
  C interfaces back into the Rust code if there is sufficient interest/need.
• Note that the API will probably change again as soon as support for more
  quadratures is added (see #43).

• Unpin version dependencies for Numpy and CFFI.

• Build warning removed.
• Add .zenodo.json.

• Do not append "numgrid" to install prefix within CMake.
• Install both shared and static libraries.
• Fortran support by default OFF.
• Python interface allows to specify the basis set instead of explicit exponent ranges (this uses https://github.com/MolSSI-BSE/basis_set_exchange).

• Grid memory management is now client-side.
• Compute one center at a time.
• Full basis set does not need to be provided.
• Separate arrays for x, y, z, and weights.
• Use Pipfile instead of requirements.txt.
• Simplify CMake structure.
• Require C++11.