I don't think << necessarily means TWO orders of magnitude so much as the less precise "LHS Much smaller then", how much smaller depends on how close you need the approximation to be.
It could be 1, 2,or 3 orders, but is it uncritical as long as R3 is a lot bigger.
I would note that there may be other constraints, so don't get silly, output drive, input bias current, thermal noise.... But the simplification in the maths works as long as R3 is much bigger then R1 || R2.
73 Dan.