doc: starpu_cuda.h needs to be included in cuda codes

doc/starpu.texi

@@ -858,6 +858,7 @@ call.
 @cartouche
 @smallexample
 #include <starpu.h>
+#include <starpu_cuda.h>
 
 static __global__ void vector_mult_cuda(float *val, unsigned n,
                                         float factor)

doc/vector_scal_cuda.texi

@@ -1,4 +1,5 @@
 #include <starpu.h>
+#include <starpu_cuda.h>
 
 static __global__ void vector_mult_cuda(float *val, unsigned n,
                                         float factor)