If somebody wants to prepare a PR I'm happy to consider this. A couple of thoughts:
It can't be the default behavior, as it will break on any number of inputs that are currently valid. To keep the code clean, it would probably be best to add an alternative alignment function that aligns by axis values. If there is code shared between the two alignment functions it may make sense to define additional helper functions.
This needs extensive unit tests to make sure it doesn't break if there are changes in how ggplot2 works. I'd like to see both vdiffr-based tests and regular tests that probe the grob structure.
This is a feature request - aligning by axis values. I came across a use case in two SO posts:
https://stackoverflow.com/q/57392541/7941188 https://stackoverflow.com/q/65550501/7941188
User Z.Lin also offered a modification of the underlining align_plots function.
Hope this is of interest for the developer. Happy new year!