any excitation voltages disturbances will affect both inputs equally,
chances are the noise will be picked up by both lines, and therefore should/will be rejected at the inputs of your differential amp. Only the true signal (i.e. the data you want),will be accepted by the diff opamp. So it has better noise immunity.
Not sure how a bridge circuit can have impedances that are close to balanced unless you employ a variable impedance resistor in the bridge? Or perhaps a ladder of resistors in series that are shorted by transistors as needed?
Yes, that's how you will need to balance your bridge (Zeroing) for the temp you want to establish as reference.
So with a single value of the potentiometer in the first leg you may compensate to zero the thermistor of ANY value in the second.
FWIW, I got that second circuit 'inspiration' from LTC with its easy drive ADCs. See this design note as an example.
Delta-sigma ADCs, with their high accuracy and high noise immunity, are ideal for directly measuring many types of sensors. Nevertheless, input sampling currents can overwhelm high source impedances or low-bandwidth, micropower signal conditioning circuits. The LTC®2484 family of delta sigma converters solves this problem by balancing the input currents, thus simplifying or eliminating the need for signal conditioning circuits.