My reference as stated was/is the SKF bearing catalogue, perhaps not as recent as your one, probably circa 1990, but neverthe less a valid design for a high speed vertical milling spindle.
If'n I have the inclination to search for it, I will post the relevant page that I photocopied at the time and since my retirement (10 years) still have.
The illustration you posted in the MRC bearing cat on page 11 bears no resemblance to the spindle design I mentioned, and I would venture to say that anyone attempting to make a spindle to that design would be treading in deep water.
I think you, and the illustration in the SKF cat you posted exagerate the "looseness" of the radial bearing, even allowing for the clearance for a lubrication film, and if'n you have a shaft with two radial bearings on it, as I recently dismantled from a cheap junk Chinese wood turning lathe, you will find as I did that it is impossible to measure even with a .0001" dial indicator any deflection of the shaft, at either end, and this is for a very basic shaft layout in the extreme...with no preloading.
BTW, a very basic bearing layout design I made for a lathe as an apprentice has been in operation since 1962, and consists of two tapered roller bearings on a shaft with 19mm bore and spaced at 100mm apart....bearing preload adjusted at the shaft end with lock nuts.
The design was by no means ideal but functional, and if'n there had been a heat problem it would have manifested itself by variation in the preload, but this configuration is very basic indeed and is not the ideal way to have a lathe spindle bearing arrangement, but due to lack of any experience in spindle design it sufficed at the time, and still functions today.....50 years later.
The "silly" statement you refer to was an "if " factor, and that has been used in a design of another machine spindle, but the bearings were in a cartridge form and not preloaded, just spaced apart to give a stiffening effect to the spindle.
Ian.