Cython iterate over list of numpy arrays without the gil

I would like to iterate over a list of numpy arrays with different dimensions, and pass them in to a cython function that does not require the GIL:

# a has T1 rows and M columns
a = np.array([[0.0, 0.1, 0.3, 0.7],
              [0.1, 0.2, 0.1, 0.6],
              [0.1, 0.2, 0.1, 0.6]])

# b has T2 rows and M columns
b = np.array([[1.0, 0.0, 0.0, 0.0],
              [0.1, 0.2, 0.1, 0.6]])

# c has T3 rows and M columns
c = np.array([[0.1, 0.0, 0.3, 0.6],
              [0.5, 0.2, 0.3, 0.0],
              [0.0, 1.0, 0.0, 0.0],
              [0.0, 0.0, 0.1, 0.0]])

array_list = [a, b, c]
N = len(array_list)

# this function works
@cython.boundscheck(False)
@cython.wraparound(False)
cdef void function_not_requiring_the_gil(double[:, ::1] arr) nogil:
    cdef int T = arr.shape[0]
    cdef int M = arr.shape[1]
    cdef int i, t

    for t in range(T):
        for i in range(M):
            # do some arbitrary thing to the array in-place
            arr[t, i] += 0.001

# issue is with this function
def function_with_loop_over_arrays(array_list, int N):
    cdef int i

    with nogil:
        for i in range(N):
            function_not_requiring_the_gil(array_list[i])

When I compile the Cython code, I get the following error:

Error compiling Cython file:
------------------------------------------------------------
...
def function_with_loop_over_arrays(array_list, int N):
    cdef int i

    with nogil:
        for i in range(N):
            function_not_requiring_the_gil(array_list[i])                                                    ^
------------------------------------------------------------

my_file.pyx:312:53: Indexing Python object not allowed without gil

Is there another type of data structure that I can use instead of a Python list to store these numpy arrays so that I can iterate over them without the gil? I'm open to suggestions involving malloc C pointers/Cython memoryviews/other types I'm not aware of.

Note that each numpy array has a different number of rows, but the length of the list of arrays is known.