Let output of `get_power` be of a special class

[jordanflitter]

I really like the SphericalPS and CylindricalPS classes that we have in tuesday. While I understand that powerbox is a more generic package that could compute power spectrum in an arbitrary number of dimensions, I think it would be useful if get_power could have a single output of some special class, rather than letting it have a varied number of outputs with different meaning according to the inputs to the function. With such an output class (that could store fields like p_k, k_bins_edges, k_centers, n_modes and var), the output of get_power would be less confusing.

[steven-murray]

Yes, I think this is something that absolutely should be done. Of course it will totally break API, so it would need to be put into v1.

The returned object should have something like the following interface:

@dataclass
class PowerSpectrum:
    power: np.ndarray
    bin_edges: np.ndarray
    bin_centres: np.ndarray
    bin_avg: np.ndarray  # the average k-mode in the bin (i.e. what is returned now when bin_ave=True)
    nsamples: np.ndarray
    variance: np.ndarray | None = None
    k_unbinned: tuple[np.ndarray, ...] | None = None  # k values in the dimensions that were not angularly averaged

And this should have relevant validation checks for the shape and type of each attribute of course.