The basic problem is that without high fidelity & a wide FOV, there's nothing that sets MagicLeap apart from any of the other companies that have been able to create AR headsets for the last ~decade.
AR right now is stuck because the display technologies aren't up to the task. The "Waveguide" approach has always fallen short - nobody's ever been able to make the viewing angle wide enough to be worthwhile, and the "passthrough" approach isn't really viable for walking down the street.
This is why I'm skeptical about Apple's AR moves - unless they've either got a totally new display technology or they've managed to do something incredibly clever with the waveguide approach, I just don't see consumer goggles working.
The only people I've seen doing something unique in this space is Tilt Five: https://www.tiltfive.com/ - they've basically done what you're suggesting: constrain the use case until you can actually build something for it.