Problem with "-f implies -n" approach is, it breaks backward compatibility. There are use cases using custom script*and* pgbench_* tables. For example the particular user wants to use the standard pgbench tables and is not satisfied with the built in scenario. I know at least one user does this way.
If we care enough about that case to attempt the vacuum anyway then we need to do something about the error message; either squelch it or check for the existence of the tables before attempting to vacuum. Since there's no way to squelch in the server logfile, I think checking for the table is the right answer.
Maybe someone can write a patch for VACUUM IF EXISTS ... :-)
