You may be chasing compounding errors.

I'd forget about the mitre slot for starters.

The table - unbolt all the wings etc then measure the table flatness. Is it still out? I'd expect the answer to be yes because if I'm picturing it correctly (I have a cheap H&F table saw) I know where the extension bits bolt on. Anyway - check and report back. What are you using for a straight edge and how do you know it's straight? How are you measuring the gap between the straight edge and the table top?

Pictures are good.

If it is as you say then using wet & dry on a sander isn't a bad way to go. That's if you only have 0.002" on the ends to remove which frankly I don't believe for an instant. I meddle in woodwork and nobody is going to notice an error from flat at that level, the wood moves more as soon as you cut it. So - total error determines suggested correction and without accurate data on this it's hard to offer useful advice.

If the error is 1-2mm then you're SOL with anything cheap and easy.

WRT the edges, shimming is better than grub screws exerting pressure. If both table edge and wing edge are exactly 90 deg according to your measurement yet they're not flat when bolted together, either your square isn't or your technique is wrong or there is dirt/burrs etc somewhere on those mating surfaces in the places you haven't checked that's causing the distortion when the edges are mated.

This sort of crap is why I have a cheap H&F saw to beat to death and an older 3 phase Woodfast saw that weighs about 5X as much but doesn't have a tilt option...... guess which one I use all the time.

FWIW I remember a friend of mine going through this with a supposedly high quality Carbatec saw some years ago so I don't doubt that the top may be out. He cut his losses & bought a more expensive saw *after* taking a straight edge and DTI with him to check it out first.

PDW