Using the name get_utf8 is not very user friendly. It is not what anyone would look for when trying to figure out how to get a string out of a bson element. To make it worse, "string" doesn't even show up in the documentation of the function, so you can't even find it when searching in the header or doxygen.

This should probably be done everywhere that utf8 is used to mean string.

You will probably need to leave the get_utf8 name available as a deprecated alias for a while.