These cars go through QC - maybe some QC folks are more slipshod than others, and the main source of lemons are the ones who were inspected by terrible QC inspectors?
Or maybe the lemons are perfectly fine cars that were damaged in shipment and were repaired badly?
Regardless, the lemon laws still protect the consumer.
It would basically turn in to a Pigovian n^3 tax on reliability.
But I suspect that you're overstating; a drive unit may well be truly independent, but something like water leaking in to the cabin wouldn't be.
Even if 80% of components have independent failure modes, the remaining 20% will account for the vast majority of your lemon law invocations.
You can easily have a couple excess failures trigger excess service, and if you're unlucky with the quality of one episode of that rework or service, end up on a never-ending rework treadmill. Even on the brands with the best quality.
The underlying assumption that justifies a category of "lemon" is that needing N repairs is evidence that it is more likely than other cars of its model year to need an N+1st repair because failures are not independent.
This may have been true of earlier technology, but it doesn't match the way modern cars are made. Mainly, the granularity of subassemblies is much higher, so for example in a Tesla they replace the entire rear powertrain or entire battery pack as a warranty item.
Lemon laws should have required some periodic statistical adjustment to determine whether lemoniness actually exists in a given brand or not.