Cycles: Add debug pass showing average number of ray bounces per pixel

Quite straightforward implementation, but still needs some work for the split
kernel. Includes both regular and split kernel implementation for that.

The pass is not exposed to the interface yet because it's currently not really
easy to have same pass listed in the menu multiple times.

intern/cycles/kernel/kernel_debug.h

@@ -19,6 +19,7 @@ CCL_NAMESPACE_BEGIN
 ccl_device_inline void debug_data_init(DebugData *debug_data)
 {
 	debug_data->num_bvh_traversal_steps = 0;
+	debug_data->num_ray_bounces = 0;
 }
 
 ccl_device_inline void kernel_write_debug_passes(KernelGlobals *kg,
@@ -33,6 +34,11 @@ ccl_device_inline void kernel_write_debug_passes(KernelGlobals *kg,
 		                        sample,
 		                        debug_data->num_bvh_traversal_steps);
 	}
+	if(flag & PASS_RAY_BOUNCES) {
+		kernel_write_pass_float(buffer + kernel_data.film.pass_ray_bounces,
+		                        sample,
+		                        debug_data->num_ray_bounces);
+	}
 }
 
 CCL_NAMESPACE_END