scipy: BUG: RBFInterpolator fails when calling it with a slice of a (1, n) array

RBFInterpolator has a bug where it throws an error when calling it with a slice of a (1, n) array. The below code reproduces the error:

import numpy as np                                       
from scipy.interpolate import RBFInterpolator            
                                                         
y = np.random.random((3, 2))                             
d = np.random.random((3,))                               
x = np.random.random((1, 3))[:, :2]                      
RBFInterpolator(y, d)(x) 

which throws this error:

Traceback (most recent call last):
  File "rbfinterp_bug.py", line 7, in <module>
    RBFInterpolator(y, d)(x)
  File "/home/hinest/miniconda3/lib/python3.8/site-packages/scipy/interpolate/_rbfinterp.py", line 416, in __call__
    out = _evaluate(
TypeError: Invalid call to pythranized function `_evaluate(float64[:, :] (reshaped), float64[:, :], str, float, int64[:, :], float64[:], float64[:], float64[:, :] (reshaped))'
Candidates are:

    - _evaluate(float[:,:], float[:,:], str, float, int[:,:], float[:], float[:], float[:,:])

The issue is that x has shape (1, 2) and stride (24, 8), which allows it to pass through x = np.asarray(x, dtype=float, order="C") at this line of RBFInterpolator.__call__ because it still is C-contiguous, but pythran throws an error saying that x is reshaped. @serge-sans-paille is this the expected behavior for pythran? If so, is there a better way to sanitize the input for pythranized functions than np.asarray with order specified?