You can see "sensitivity" plots for the various methods we have for observing gravitational waves. The horizontal axis is usually in terms of the frequency of t

You can see "sensitivity" plots for the various methods we have for observing gravitational waves. The horizontal axis is usually in terms of the frequency of the wave and the vertical axis is how sensitive we are to them. This article is about "nanohertz" waves, which means their wavelength is roughly 10⁹sec × speed of light, i.e. 30ish light years. So, it seems we can detect really "big" gravitational waves. If you look through the aforementioned plots you will see that sensitivity drops off significantly for "more astronomical" wavelengths, so there are definitely scales at which we can not detect, as you suggested. There might be some interesting ways to introduce very long gravitational waves as solution of current discrepancies in our understanding of cosmology, but they are also probably introducing more discrepancies than they solve. I would guess people have considered these ideas, because "cosmological scale general relativity" is pretty heavily researched and mainstream.