However, I can see a scenario where, despite that, it is not the right choice for Tesla. It's important to remember that self-driving technology will be commodity at some point (like the lidar units themselves). Whoever offers the most affordable option meeting some baseline performance will dominate the market and licence to every self-driving fleet. It wouldn't surprise me if Waymo ended up meeting that criteria first and licensing the tech to everyone from Uber to Toyota.
Once that tech is out, anyone else that succeeds in developing comparable tech effectively has an uphill battle ahead of them trying to sell their new, less proven version at a price and volume that can recoup their R&D investment. I think we'll see a lot of players leave the race once the first licensable product hits the market. In fact, I think we've seen a bit of that already.
Which brings me back to Tesla. My guess is their strategy is not just to develop self-driving - they'd lose on equal ground (due to not being first) - but to develop "2nd gen" self-driving which has a specific competitive advantage of not requiring lidar and thus having lower total cost per unit. This allows them to enter the market late but still have a significant edge in that market.
If a company lets their engineer's use lidar to develop self-driving tech, then the tech they build will be 100% dependent on lidar. That's Parkinson's law. If, on the other hand, the company adds a strict constraint that lidar is not an option, then the engineers might fail to build something that works - or else they will produce something with a unique advantage.