Firstly welcome to the forums. Secondly as somebody who did years of CNC machining quality control (even though I’ve only had a Sovol printer at home about a month), there are two completely different measurements here, levelness and flatness, and different controls for each.
Levelness wise, the gantry can handle >10mm static adjustment before binding. The automatic process “Quad-Gantry-Leveling” does this before (or even without) a bed mesh. This is not a question of variations in aluminum that lives under the adhesive magnets, under the PEI coated steel working surface the eddy sensor reads, just where the 4 lead screws are vs each other. Far from a bandaid in this regard, what you have is a better solution than using manual screws to tilt the bed to match the gantry, and isn’t nearly as easy to upset by doing things like yanking parts off the bed before they are quite cool enough as with older designs. A sag on one side of that much should be removed during this step.
Flatness wise, flat is simple, until it isn’t. Your readings will change considerably between probing cold and probing a 100’c bed. This difference in thermal expansion between materials and compared to the mounts is the primary reason most people see the Taco btw, not manufacture quality. Taking the PEI sheet off and reapplying after it heats will also cause changes by relieving some of those stresses, but while cool for demonstrating the concept, its not something I can personally be bothered doing daily when the automatic software solution exists.Milling the bed could only make it correct for one of these possible case, and it’d be super sensitive to the assembly and shipping caused differences so NO major brand does this as a solution on a current production model. This is instead the measure where bed mesh is designed to solve (even in a many hundred thousand dollar mill, it will touch off the part in many places after you clamp it down, not trust the machined bed), if you’ve set up the previous step correctly, you’ll note four corners that will be zero on the mesh, with adjustments for the flatness (not level) between them.
IF you’ve followed all the manuals steps to the letter and its still not coming quite low enough displayed values in the mesh for your tastes, consider also that twist in the machine by the surface its placed on can be a factor, your table or counter is not a dead even surface plate and likely has designed fall or daft angle too it. Try putting a coin under one foot, and without changing gantry quad level, rerun the bed mesh to see how that works. You cannot build a machine strong enough to avoid this, I’ve seen bridgeport mills that weigh hundreds of kilos of steel on a fairly narrow pedestal and still have a working area barely bigger than the SV08 require shims under one corner to compensate for the shape of the concrete floor, and larger routers come on screw jack feet to wind the frame up tens of millimeters to find a even level over their many meter spans.
So yeah, there isn’t a “fix at the hardware level” answer to be had here that isn’t already implemented on the SV08. If you think this sounds like shilling, its not, it applies to ALL the major brands. You will have the same process in a Prusa, Qidi or Bambu large bed coreXY machines, indeed the only difference is some move the bed with 3 points contact, others the gantry with 4 points contact. But the same process is done with the same probes. People are using that complete list of different printers including the SV08 and getting amazing results on large prints.. so if you personally are having issues getting it to stick in a corner, great we know a starting place, lets troubleshoot that because I’m sure you can get a result you are very happy with, as others already are on the same geometry. PLA not sticking on one side isn’t normal for the design and if caused by bed alignment alone indicates a height error FAR more than bed-mesh should allow to exist.