You can build accurate 10:1 dividers yourself using the 3 matched Rs in parallel and matched to another Rb then the the three Rs in series is 9 times Rb so Rb is 1/10 the total resistance.
This means you might get away with one expensive 0.01% resistor, a 10:1 divider then use it to make the other ones yourself (using trimmers).
You can use your meter to do the matching as it doesn't need to be calibrated to match Rs, just give consistent readings (or you could use a bridge circuit).
A further thought, if you matched a second resistor to Rb then you could have a bridge with (3Rs+Rb) in series with the second Rb on one side, the known R (0.01%) in series with the one you're making on the other and trim until balanced. You can thus go from R to R/10 and so on though I guess errors would mount.
To get around the cumulative errors, or at least get a check, you could perhaps get two 0.01% resistors one at each end of the range then you could work in both directions and compare your final home-made with the known good value.