You should do the math here. The matching needs to be _stable_ if you're using a bridge, but _accuracy_ of matching is only important in reducing errors from large lead wire resistances (assuming zero/span are trimmed or calibrated). You shouldn't have large lead wire resistances, particularly with the 3-wire configuration, since they are not guaranteed to be tightly matched to begin with. Use higher RTD resistance, fatter wire, shorter runs or Kelvin connection.. whatever it takes. It would be an act of faith to assume wires in a bundle are matched to even 1%.
Of course, but it will be quite expensive, you certainly don't need it to meet your spec. All you need to get 0.05% matching is tightly matched resistors, low offset op-amps and a pass device that uses only leakage control current (MOSFET or JFET). Even high beta BJTs are almost good enough if they're sorted into beta bins, use a Darlington and they're fine. A resistor pair matched to 0.05% is about $5 one-off, 0.01% individual parts (0.02% matching) will run you about four times that. The semis are cheap.