Collisions aren't theoretical, near misses are so common that there's an entire department at NASA dedicated to detecting them and warning satellite owners to adjust course, I know because we were contacted about a possible collision involving our cubesat. Prior to megaconstellations being deployed if humanity stopped adjusting satellite orbits there would be a collision within a month, now there would be a collision within 5 days. It's only a matter of time until both satellites on a collision course don't have the ability to adjust course (engine failure or no propulsion/fuel/comms). In the event of a Carrington-style solar flare there's a good chance a decent percentage of satellites would be knocked out, making this hypothetical into a reality. Further, we can only currently track objects down to about 10cm, but NASA estimates suggest about 500,000 objects exist between 1-10cm in size in LEO.

Starlink satellites aren't large enough to survive the heat of reentry. A more likely concern is the various materials vaporizing and dispersing into the atmosphere, as was mentioned in the article.

That being said, calling them "heavy metals" like the interviewee did is rather dubious. We're not talking about lead, as what most readers imagine when they hear that term. It's mainly aluminum and copper. The person interviewed is picking their words to overexaggerate their claims

Well if the atmosphere decides it's going to have a day off then I'll start to worry.

The temperatures generated by reentry are not just hot enough to vaporize a satellite they are way beyond hot enough to vaporize a satellite. I can't imagine any scenario where a satellite survives. In any case the vast majority of the orbits are controlled, Which means they come down over the ocean.