Momentum is only conserved in aggregate when there is no external influence. Parts of the system can still transfer momentum between each other. It's quite permitted for us to change our momentum, as long as other things have a change in momentum that is equal and opposite to our change.

I don't have specific sources for "net momentum of the universe is zero in the CMB frame of reference", but it sounds like a good expectation. If we found a way to test that, it would provide either a confirmation or open new scientific questions. Unfortunately, as the observable universe is only a subset of the whole universe, I suspect that we cannot determine the net momentum of the whole universe.

The momentum of the Earth, or a vehicle, or the Sun, or our galaxy, could vary from the aggregate due to any number of interactions. For instance, the Earth orbiting the Sun must involve the Earth and Sun having different velocities, so they can't both match the CMB. As it happens, neither does.

Even pointing a flashlight into space would cause a tiny change of momentum for the Earth: the outgoing photons have momentum, so the flashlight (hence the Earth, indirectly) must experience a change in momentum in the opposite direction.