There are a lot of different things called "Lidar" : at a minimum, Structured Light, Time of Flight, and triangulation are the three most common methods. Within each method, there are vastly different approaches - scanning vs. single-frame, frequency modulated vs. pulsed, and more.
Each approach comes with different trade-offs suited to different use-cases. In the case of the iPhone rear camera, I believe the approach used is single-frame pulsed Lidar using a VCSEL (vertical cavity surface-emitting laser), a dot-grid diffuser, and SPAD (single photon avalanche diode) CMOS camera.
Basically, a shifting grid of dots at a very specific wavelength is projected outwards from the laser, and an unbelievably, photon-level sensitive camera tuned to that precise wavelength measures how long each dot takes to come back after it's emitted. In this way, Time of Flight can be measured and a precise distance determined.
This system provides incredibly high resolution at a moderate distance (in this case, 5m or so) at a moderate cost.
For a car, this system wouldn't be very plausible due to range, resolution, and output power constraints. The size of laser required to emit a coherent dot matrix which could reflect at a long distance would grow to an extreme degree, the noise floor of the SPAD would eventually render the system useless, and the amount of data required to produce a higher resolution dot cloud would become quite challenging, as these systems generally need to be heavily filtered and sub-sampled, so an enormous amount of input data processing is required to produce a reasonable output.
The main reason that Tesla don't use Lidar is that they decided to bet on optical (IMO, foolish), but the reason they decided to bet on optical is complex and definitely in no small way cost driven. They couldn't just slap an array of iPhone sensors (I believe they're Sony-manufactured) on a Tesla and call it good.