I mean, from 4.16 to 4.17 for example, and if LTS is instaled from one kernel to the new LTS ones. So the change is transparent for the user. Only to be installed by the user and not automatically so no problems there. But I think is a good thing to have it, I thing that a year or so before was avalaible in repositories like linux-latest and linux-lts-latest or something like this. Why to remove it?
The topic has been discussed countless times over the years.
At one time there was different packages doing what you request but they caused more trouble than worth the effort.
A kernel change is a user decision - there is no way to automate this in any productive and failsafe manner.
Being optional and been advised to be experimental is not sufficient?
Then you create a custom package - which you will need to maintain as needed - you can use the abandoned PKGBUILD as template - just don’t complain of the consequences.
Just remember - there’s is a reason they were abandoned.
We’re already getting blamed for upstream decisions.
An experimental tool people use and then complain about because they didn’t read the warning? No thanks.
What does upstream have to do with this though?