[CFL] Change cfl_load to use width and height

Since the size used with cfl_load can either be based on the transform
block size and the prediction block size, width and height are used as
parameters instead of TX_SIZE.

This resolves a problem where cfl_compute_alpha_ind was reading
uninitialized memory.

Change-Id: I187dbdd5b2e8bd85e82bb77eb74859bee2cd3f1e
diff --git a/av1/encoder/encodemb.c b/av1/encoder/encodemb.c
index e22b60c..adb8eac 100644
--- a/av1/encoder/encodemb.c
+++ b/av1/encoder/encodemb.c
@@ -1581,7 +1581,7 @@
   uint8_t tmp_pix[MAX_SB_SQUARE];
   // Load CfL Prediction over the entire block
   const double y_avg =
-      cfl_load(cfl, tmp_pix, MAX_SB_SIZE, 0, 0, max_txsize_lookup[bsize]);
+      cfl_load(cfl, tmp_pix, MAX_SB_SIZE, 0, 0, block_width, block_height);
 
   int dist_u, dist_v;
   int dist_u_neg, dist_v_neg;