Remove uint64_t cast for gridCols * gridRows

Since gridCols <= 256 and gridRows <= 256, the gridCols * gridRows
product can be represented in uint32_t, so the uint64_t cast is not
necessary.

Note: We can alternatively test (gridCols == 256) && (gridRows == 256).

src/write.c

diff --git a/src/write.c b/src/write.c
index 50a451e..5266ff3 100644
--- a/src/write.c
+++ b/src/write.c
@@ -2154,7 +2154,7 @@
     if ((gridCols == 0) || (gridCols > 256) || (gridRows == 0) || (gridRows > 256)) {
         return AVIF_RESULT_INVALID_IMAGE_GRID;
     }
-    if ((uint64_t)gridCols * gridRows > 65535) {
+    if (gridCols * gridRows > 65535) {
         // Section 8.11.12.1 of ISO/IEC 14496-12: "All the references for one item of a specific type
         // are collected into a single item type reference box", whose reference_count field is
         // unsigned int(16) whatever the 'iref' version (Section 8.11.12.2). Moreover, ISO/IEC 23008-12