And also use mm_base_modem_at_command() instead of the full() version, as we're anyway mm_base_modem_peek_best_at_port(). This commit also fixes the logic where we make sure the finish() method completes the GTask that is created in the async method originally.