radv: set the correct buffer size for NGG streamout

If the size is passed through CmdBindTransformFeedback() uses that.

This partially fixes dEQP-VK.transform_feedback.simple.multiquery_1
by reporting the correct number of primitives written (the computation
is based on the buffer size). There is still a bug around GDS offsets
that will be fixed later.

Tested on GFX10.3 by forcing NGG streamout.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19433>

src/amd/vulkan/radv_cmd_buffer.c

@@ -4118,12 +4118,11 @@ radv_flush_streamout_descriptors(struct radv_cmd_buffer *cmd_buffer)
           */
          uint32_t size = 0xffffffff;
 
-         /* Compute the correct buffer size for NGG streamout
-          * because it's used to determine the max emit per
-          * buffer.
+         /* Set the correct buffer size for NGG streamout because it's used to determine the max
+          * emit per buffer.
           */
          if (cmd_buffer->device->physical_device->use_ngg_streamout)
-            size = buffer->vk.size - sb[i].offset;
+            size = sb[i].size;
 
          uint32_t rsrc_word3 =
             S_008F0C_DST_SEL_X(V_008F0C_SQ_SEL_X) | S_008F0C_DST_SEL_Y(V_008F0C_SQ_SEL_Y) |