[CFL] Fix int type in padding function

Int was used but should have been int16_t. This does not alter the
encoder/decoder behavior

Change-Id: I6bfd8a3942e9f9b86cccf10c144fec147e996944
diff --git a/av1/common/cfl.c b/av1/common/cfl.c
index cb0e227..9ffaae9 100644
--- a/av1/common/cfl.c
+++ b/av1/common/cfl.c
@@ -46,7 +46,7 @@
     const int min_height = height - diff_height;
     int16_t *pred_buf_q3 = cfl->pred_buf_q3 + (width - diff_width);
     for (int j = 0; j < min_height; j++) {
-      const int last_pixel = pred_buf_q3[-1];
+      const int16_t last_pixel = pred_buf_q3[-1];
       for (int i = 0; i < diff_width; i++) {
         pred_buf_q3[i] = last_pixel;
       }