There are only very very few examples where this can be helpful. But in very very most cases such a filter would be contradictory! Remember what we want to accomplish by grounding and shielding: To force all ground potentials, where ever they are, to show the SAME value. So, introducing impedances between different ground points can even increase the ground noise, because then charge injection due to stray capacitance (think of intrinsic space charge capacitance of all bodies!) will not find a low ohmic path for the ground return current and will result in a rise of separated ground potentials.

But you are right in the following: We need to prevent that such charge injections via stray capacitances will occur at all, because voltage drops of according ground return currents will be troublesome in any case, means will increase ground noise, or to be more precise, common mode noise. And it's just the task of shielding to prevent this: If a grounded shield between separated ground points is present, charge injection will end at the shield and LOCALLY flow back to its ground, so that ground return currents BETWEEN separated grounds are heavily decreased!
Another way to prevent these ground return currents is to introduce common mode filters. They do the same, namely decrease the amount of ground return currents BETWEEN separated grounds.
Common mode filtering shows the best results if symmetrical signal routing is used. Then, if the symmetry is kept high, there's no ground return current just by design!

Concluding: Common mode noise is kept under control, if TWO factors are fullfilled:

1. Minimizing ground impedance between all involved ground points by the use of wide and thick (solid) ground planes (radio frequency planes), and

2. Minimizing of ground return currents flowing BETWEEN separated ground points by intelligent signal routing (use of symmetrical signal routing), by using common mode filters and by using proper shielding (Faraday cage sections).

Kai