Welcome to the rabbit hole, please enjoy your journey and don’t miss tea with the white rabbit whilst you are here 
I may also have fallen in this hole, here’s the post that resulted from my trip down measuring lane Backlash, Deflection and Vibration
It’s true, you can always make some improvement to a machine or an operating condition. If we were able to mount our Shapeoko to a rigid flat, well damped surface with a similar thermal expansion coefficient to the machine frame, we’d be able to largely eliminate frame and spoilboard deflections. However, it’s not a mill, it’s a gantry router and there’s a reducing reward for additional effort in many of these things.
In the case of the Pro, when I did the deflection calculations for the old SO3, the X beam section, for a 5kg force applied in the centre (the most sensitive point) should show a deflection of < 0.025mm. In the case of the Y axes that would be ~ 0.025mm if the HDZ and spindle are both at one end of the X rail, half that if the Z is in the centre.
I personally have my old Shapeoko 3 XXL bolted down to a large torsion box base in the 4 corners, with rubber damping under the spoilboard to keep the working surface flat and damped from vibration. That is much less of an issue on the new machines with the much better designed frame.
I’d be more concerned about the spoilboard moving than the machine gantries under any ‘normal’ cutting loads, quite a few people have had good success putting rubber floor mats or similar vibration absorbing layers under the machine. This allows the machine frame to move with thermal expansion & contraction & find it’s own level, whilst supporting and damping the work area.
The level spoilboard issue is a maintanability issue too, even once levelled once, unless the whole area is attached to a much stronger frame it may well move between levelling and machining the actual workpiece, or be flexed by the workpiece under clamping forces. If you’re going here, the spoilboard needs fixing down, not just resting on a flat thing.
As for metal cutting and large surface blocks…
I did repeatedly consider buying an old iron surface table to install the machine on top of from The Surface Table Company seller on eBay UK. Then I realised that renting the truck with the fork lift to pick it up and get it into my workshop might cause some questions at home…
I also measured that the deflections are notably smaller in the corners of the machine than the centre, so, unless your metal part is 4ft by 4ft, stick it in a corner, your gantry extrusion deflection will be minimised by geometry and your ballscrew deflection is basically constant across the length of the axis.
HTH