Move a memoryview test out of the "run" test directory since memoryview tests...

Move a memoryview test out of the "run" test directory since memoryview tests tend to be slow and have their own test directory.

tests/memoryview/cfunc_convert_with_memoryview.pyx

+# mode: run
+# tag: autowrap
+# cython: always_allow_keywords=True
+
+cdef void memoryview_func_a(double [:] x):
+    x[0] = 1
+
+cdef void memoryview_func_b(double [::1] x):
+    x[0] = 2
+
+cdef void memoryview_func_c(int [:] x):
+    x[0] = 1
+
+cdef void memoryview_func_d(int [:] x):
+    x[0] = 2
+
+cdef void memoryview_func_e(int [:,::1] x):
+    x[0,0] = 4
+
+cdef void memoryview_func_f(int [::1,:] x):
+    x[0,0] = 4
+
+
+def test_memview_wrapping():
+    """
+    We're mainly concerned that the code compiles without the names clashing
+    >>> test_memview_wrapping()
+    1.0
+    2.0
+    1
+    2
+    """
+    cdef a = memoryview_func_a
+    cdef b = memoryview_func_b
+    cdef c = memoryview_func_c
+    cdef d = memoryview_func_d
+    cdef e = memoryview_func_e
+    cdef f = memoryview_func_f
+    cdef double[1] double_arr = [0]
+    cdef int[1] int_arr = [0]
+
+    a(<double[:1]> double_arr)
+    print(double_arr[0])
+    b(<double[:1:1]> double_arr)
+    print(double_arr[0])
+    c(<int[:1]> int_arr)
+    print(int_arr[0])
+    d(<int[:1:1]> int_arr)
+    print(int_arr[0])
+    # don't call e and f because it's harder without needing extra dependencies
+    # it's mostly a compile test for them

tests/run/cfunc_convert.pyx

 # mode: run
+# tag: autowrap
 # cython: always_allow_keywords=True
 
 cimport cython
@@ -229,51 +230,3 @@ def test_cdef_class_params(a, b):
     TypeError: Argument 'b' has incorrect type (expected cfunc_convert.B, got cfunc_convert.A)
     """
     return (<object>test_cdef_class_params_cfunc)(a, b)
-
-cdef void memoryview_func_a(double [:] x):
-    x[0] = 1
-
-cdef void memoryview_func_b(double [::1] x):
-    x[0] = 2
-
-cdef void memoryview_func_c(int [:] x):
-    x[0] = 1
-
-cdef void memoryview_func_d(int [:] x):
-    x[0] = 2
-
-cdef void memoryview_func_e(int [:,::1] x):
-    x[0,0] = 4
-
-cdef void memoryview_func_f(int [::1,:] x):
-    x[0,0] = 4
-
-
-def test_memview_wrapping():
-    """
-    We're mainly concerned that the code compiles without the names clashing
-    >>> test_memview_wrapping()
-    1.0
-    2.0
-    1
-    2
-    """
-    cdef a = memoryview_func_a
-    cdef b = memoryview_func_b
-    cdef c = memoryview_func_c
-    cdef d = memoryview_func_d
-    cdef e = memoryview_func_e
-    cdef f = memoryview_func_f
-    cdef double[1] double_arr = [0]
-    cdef int[1] int_arr = [0]
-
-    a(<double[:1]>double_arr)
-    print(double_arr[0])
-    b(<double[:1:1]>double_arr)
-    print(double_arr[0])
-    c(<int[:1]>int_arr)
-    print(int_arr[0])
-    d(<int[:1:1]>int_arr)
-    print(int_arr[0])
-    # don't call e and f because it's harder without needing extra dependencies
-    # it's mostly a compile test for them