variable inherit.aes behavior for geom_vline and friends

[EvaMaeRey]

In #6633 the inherit.aes behavior is exposed to the user, but the default is set to FALSE, which means that that layer is not in sync with most user-facing geom_* layers. I think setting this to NULL (so still exposing to user), and turning inheritance off if xintercept is non-null would feel more consistent when this geom_* family is used in a non-annotative way. Thoughts?

[teunbrand]

It is not that fresh in memory anymore, but I think the default is set to FALSE to preserve backward compatibility. As long as that is preserved, I have no qualms with the proposed behaviour.

[EvaMaeRey]

Great! I've sometimes thought that x could/should be an alternate for the xintercept aesthetic for geom_vline(), and y for yintercept... (but leaving it so setting xintercept and yintercept triggers annotation behavior). But maybe there is not analogue for geom_abline. Still I wonder if there are reasons for not pursuing those changes. I'll put aside for now to focus on the inherit.aes questions!

[EvaMaeRey]

It feels like the geom_abline is a little different in kind because of the slope, intercept defaults (1 and 0 if not provided) whereas this doesn't happen for geom_hline and geom_vline with xintercept and yintercept. Perhaps inherit.aes needs to stay FALSE given these internally provided defaults for the geom_abline case - will give it some more thought!