Apologies, I got 2000km confused with another megaconstellation, later rollouts / V2s+ are suppose to be up to 1200km, which was initially filed / granted with FCC. They did request/allow to move some of of larger v2s to lower orbits, but the full megaconstellation plan won't be constrained to <350km simply because there aren't enough orbit slots (as managed by UN/ITU) for the constellation size star link envisions. Below is recent image of current starlink distribution. Most are 400-500km and above, i.e. much longer decay times. My understanding is they're throwing v2 "minis" which still weight 3x more to lowerish orbits because that's most economical for F9 delivery, but once they have more payload via starship, full size v2+ is going 500km-1200km. 500km more altitude as like ~4 milliseconds of latency, which is not nothing, but still minor vs economic benefits of more coverage with less hardware. IMO current low LEO focus isn't ... starlink being responsible, it's result of cost optimization of coverage:payload for F9. Starship will come with different set of cost optimizations, likely for higher orbits using larger hardware, but less of it.
AFAIK Starlink does not plan to put satellites in 1200 km orbits. All Starlink satellites are in orbits of 600 km or less, where any debris naturally decays in less than 5 years:
> Starlink satellites operate in a low Earth orbit below 600 km altitude. Atmospheric drag at these altitudes will deorbit a satellite naturally in 5 years or less, depending on the altitude and satellite design, should one fail on orbit. SpaceX proactively deorbits satellites that are identified to be at an elevated risk of becoming non-maneuverable. This proactive approach minimizes the number of non-maneuverable satellites in space.
https://imgur.com/a/VuweZZo