In doing some of the conversion to make expert info "items" filterable it seems there's need for a "proto_tree_add_expert" function. It would solve the cases of when "proto_tree_add_text + expert_add_info" or "proto_tree_add_none_format + expert_add_info" are used, but what's really desired is "expert info with byte highlighting"
The expert info API functions all end up calling expert_set_info_vformat, a static function within expert.c. expert_set_info_vformat makes all of the necessary proto_tree_add_* calls to populate expert info. Does it make sense to add proto_tree_add_expert to expert.h instead of proto.h so that expert_set_info_vformat doesn't need to be exposed? I'm trying to figure out the best way to encapsulate such functionality.