Lock-free compaction. Why not?

So anyways I talked last week about lock-free vacuum. Since then I almost finished creating a plugin that works on all Jetbrains products. It hooks to the internal database tool and displays the internals of the database and some stats. I would use this for my next phase: autovacuum with compaction.
The thing is, after reading the code a million times, I still don't understand why lock-free (or minimum locking) is such a big problem! Is it that hard to lazily move tuples from one page to the other after defragmenting it lazily?
I even made (a super simple equation) to calculate how much disk space can be reclaimed, and created a full list of the advantages including maintaining clustered indexes. This would of course be followed by benchmarks.
Here is the wip plugin btw https://github.com/ahmedyarub/database_internals_plugin/tree/2.0.0 which should be released this week.
Like I don't even trust ChatGPT at all but I kept on trying to find reasons for not doing it but I couldn't find any!
Such a trivial change that can bring a huge advantage. I just hope that somebody could find me a reason for why it wouldn't work.