The laser scan serves two purposes in navigation:
It is used by the costmap to detect obstacles so the robot doesn't hit them. If you have no obstacles that aren't on your map, or if you have some other way of avoiding obstacles, you do not need the laser scan.
It is used by AMCL to detect where we really are on the map. It provides corrections between the odom frame, which is (roughly) the integration of the cmd_vel over time, and the map frame. If your cmd_vel integration is perfect and doesn't drift, you do not need AMCL and do not need the laser scan. This might work in simulation, but probably won't work in the real world with imperfect sensors and imperfect robot motion. Overtime, the integration and the real-world are certain to diverge.
However, for testing purposes, if you want to try this, I think there is nothing you need to do except make sure you have something publishing the map->odom transform. AMCL should not interfere; if there are no /scan messages, it shouldn't publish any transforms, I believe. And costmap will still work if no scans are received.
In the long run, if you want to continue to run this way, you should make a custom launch file based on the nav2_bringup/localization_launch.py that does not launch AMCL and change the costmap parameters, so it does not use the obstacle layer.
