Generic numbers for dimension sizes

[hixan]

Could generics be implemented for the specific size of a dimension? or perhaps even generics for multiple dimensions at once?

An example of where this may be useful is in a covariance matrix function:

def var_covar(data: NDArray[T, Any], mean: NDArray[T]) -> NDArray[T, T]:
    '''calculate variance-covariance matrix with prior mean'''
    diff = X - mean
    return (diff.T @ diff) / X.shape[0]  # X.shape[0] will always have the same value as T