Numpy - How to filter a 3d array for only first non-repetitive block?

This question is an extension to my original post earlier

Numpy - How to filter and shorten a 3d array?

Suppose I have the following 6 by 3 by 3 array

array([[[-2, -2, -2],
        [-2, -2, -2],
        [-2, -2, -2]],

       [[-2, -2, -2],
        [-2, -2, -2],
        [-2, -2, -2]],

       [[-2, -2, 71],
        [-1, -1, -1],
        [71, -1, 52]],

       [[-2, -2, -2],
        [-2, -2, -2],
        [-2, -2, -2]],

       [[-2, -2, -2],
        [-2, -1, -2],
        [74, 78, -2]]

       [[-2, -2, -2],
        [-2, -2, -2],
        [-2, -2, -2]]])

Now I would like to filter such array by the following standards:

Treat each 3 by 3 array as a block, if all elements in this block = -2, we take this block as 'empty', we should only cut the block which are consecutively 'empty' from the start and from the end until there occurs a non-'empty' block, so the target array would look like this(3 by 3 by 3):

array([[[-2, -2, 71],
        [-1, -1, -1],
        [71, -1, 52]],

       [[-2, -2, -2],
        [-2, -2, -2],
        [-2, -2, -2]],

       [[-2, -2, -2],
        [-2, -1, -2],
        [74, 78, -2]]])

You can reproduce the array by

array = np.array([[-2,-2,-2,-2,-2,-2,-2,-2,-2],
    [-2,-2,-2,-2,-2,-2,-2,-2,-2],
    [-2,-2,71,-1,-1,-1,71,-1,52],
    [-2,-2,-2,-2,-2,-2,-2,-2,-2],
    [-2,-2,83,-2,-76,-2,-87,-2,-2],
    [-2,-2,-2,-2,-2,-2,-2,-2,-2]])
newarr = array.reshape(6,3,3)