Use pattern variables for instanceof for type casting in equals

[arteam]

Leverage the JEP 394 and express type casting and the equality check as a single expression.

The biggest issue seems to be that using instanceof instead of getClass actually can change the semantic of equals for non-final classes: two different children of the same parent can be equal if they don't override parent's equals.

[elasticmachine]

Pinging @elastic/es-core-infra (Team:Core/Infra)

[pugnascotia]

The biggest issue seems to be that using instanceof instead of getClass actually can change the semantic of equals for non-final classes: two different children of the same parent can be equal if they don't override parent's equals.

How concerning is this? Should we be being this to Core/Infra for discussion, before we change 800+ classes? I do think this PR is too big to review. If we want to proceed, we should break this down into smaller chunks.

[arteam]

How concerning is this? Should we be being this to Core/Infra for discussion, before we change 800+ classes? I do think this PR is too big to review. If we want to proceed, we should break this down into smaller chunks.

I'm okay to breaking this PR in multiple chunks. I believe it's not very concerning since all the "abstract" classes have children which override the equals method. But I can specifically extract such classes from the change. I can find 101 matches for super.equals(obj) == false in the code base, so there should be around 20-30 abstract classes provided that a base class has a handful amount of children.

[henningandersen]

I would prefer not to do this, for two reasons:

1. equals is then no longer symmetric. Suppose you have class B extends class A and corresponding objects b and a. If you do a.equals(b), it could be true, whereas b.equals(a) would not be.
2. The equals generated by IDEA differs from this. We thus have to hand-craft equals, which is error-prone.

[arteam]

Thank you Henning!

I actually was going to close the PR exactly for these both reasons. I found that there are too many side effects for going from getClass to instanceof (especially for parent classes) and also because Intellij doesn't provide a template for generating equals with instanceof for non-final classes. Anyways, this PR was a very useful exercise for me for learning some advanced refactoring capabilities of IntelliJ.