quaternion

Adds a dedicated quaternion dtype, allowing quaternions to be used as first-class citizens in NumPy arrays, as demonstrated in the basic usage examples with direct array operations.

Implements core ufuncs like add, multiply, exp, and log for fast operations on quaternion arrays, enabling efficient batch processing without Python loops.

Quaternion arrays can be viewed as arrays of doubles without copying data, as shown with as_float_array, ensuring seamless interoperability and minimal memory overhead.

Provides utilities to convert between quaternions, rotation matrices, vectors, and Euler angles, though Euler angles are minimally supported due to their noted drawbacks.