I'm concerned that this a public class. Even though it's in a package with .internal. in the name, somebody out there is probably using it. Maybe just mark it deprecated?

Marking it deprecated for now would also be alright for me, but I think it should be removed eventually. There is no point in including a class (and maintaining tests for it) which is not used by Gson at all; if we ever need it again we can revive it from the Git history. It appears that class was originally introduced to protect against DoS attacks (see this StackOverflow comment), but Java's HashMap implementation has been adjusted for a quite a while now already to protect against hash collision DoS attacks (see JDK-8046170), so if someone is using Gson's LinkedHashTreeMap they can probably just switch to using Java's LinkedHashMap (which extends HashMap) or TreeMap. Maybe there is even no need for Gson's LinkedTreeMap anymore and Gson could use Java's LinkedHashMap instead, but that is a different story.
What do you think?

Users use classes from internal at their own risk, we should merge it.

OK, given that the need for this class is obviated by DoS protection in the JDK and given that the class is .internal. I think we can probably delete it. However, as I mentioned in #1989, I'd like to make a Gson release before merging these potentially-breaking changes.

2.8.9 has now been released, so as agreed I am merging this.