OK, lets do a sampled data SpaceTime expansion

Vacuum bubbles expanding, sample rate dropping. Why? Either momentum left over from the bang, or it is adjusting to light passage. Source and destination are already normalized, it is only interstellar space that has the problem.

So this is the same model, except we have Nyquist to guide us. Light starts out blue, it is sampled faster. Along the way, the sample rate slows. What happens? Beam spread. under sampling is spacial.  What is the difference? My space impedance adjusts along with the vacuum.  My model, as a vacuum sampled data system, has no other properties, except sampling everywhere, at about the same rate, producing the same hyperbolic spread of phase imbalance as required by phase minimization. The effect shows up in any line of symmetry the observer can view.

We solve it the same way, take one wave of light, at the source, and assume the expansion is isotropic, count up the vacuum expansion, proportional to r cubed scaled by the reduced sample rate.  Wait! times the sample rate? Either that or time. For any local observer, the sample rate is the same, with respect to matter. So it matters not.  What matters is the r cubed, I include it, expansion in all directions. and the light wave is not assumed to be ray, infinitely small.

Wait, you say, the wave spread is  minimal transverse to the direction of travel.  Maybe it is. But it has to be minimal with relative to your aperture. And you have to assume that space impedance is isotropic relative to the dropping sample rate. Then you have to account for lower x,y bandwidth on your focal plane, and make sure the lower bandwidth is less than the red shift you are measuring.

Collecting light for longer intervals does not over come  lower transverse bandwidth loss of your focal plane, unless you assume light is quantized, and can be treated as a small diameter ray. But then the effect of spacetime expansion on space impedance is gone. If you want to eliminate the space impedance problem, then you have to show the vacuum has enought SNR to perform the
sin(theta) = theta calculation to the third power. The vacuum normally does that operation by renormalizing the beam over some surface area of the wave front. That is a space impedance function.

OK, now gravity wave.

Gravity cannot really outrun light.

With exponentially expanding space, two nearby observers are separated very quickly; so much so, that the distance between them quickly exceeds the limits of communications. The spatial slices are expanding very fast to cover huge volumes. Things are constantly moving beyond the cosmological horizon, which is a fixed distance away, and everything becomes homogeneous very quickly.

This can heppen, a huge jump in the volume of space, at the lowering sample rate of space. But you end up with a wave crest, a high sample on one side, a low sample on the other.  I am not sure what happens. If not, then you have to speculate on something other than vacuum. But we are running out of 'order' Higgs at 91 (wave), the vacuum at zero, maybe four accounting for noise. I dunno what is beneath.