feature(ecs-gestalt): Migrate Components to gestalt's Components.

[DarkWeird]

Contains

This is the first step of re-integration gestalt's Entity System (gestalt-entity-system) to Terasology by switching to use gestalt's Component interface.

This PR does not yet introduce any entity system logic from gestalt, it just switches out Terasology's Component marker interface for the one from gestalt.

As gestalt assumes that Component#copyFrom is doing a deep copy, a potential bug is non-cloning collections and nested types properly at component data copy. Since we're not yet using this functionality we have to keep this pitfall in mind once we're switching to use gestalt-entity-system completely.

How to test

All branches for this migration PR are named feature/migrate-ecs-to-gestalt.

for m in . $(ls -d modules/*/); do                                                                                                                    
	git -C $m remote update -p
	git -C $m switch feature/migrate-ecs-to-gestalt
	git -C $m merge --ff-only @{u}
done

• Compile (Ωmega workspace)
  • gradlew compileJava
  • gradlew compileTestJava
• Run Game
• Start Game
• Run Tests
  • gradlew test

Need ToDo

• Merge backport(v7): refactor!: rename Component::copy to Component::copyFrom (#122) gestalt#123
• Migrate omega module space (see PRs referencing this PR)

[skaldarnar]

I like that you try to push adoption here now that we have the foundation based on gestalt v7 👍

As mentioned on Discord, I'm looking at the code right now and I'm wondering about the "deep copy" advise given in the gestalt wiki.
We have lots of components holding collections of elements, and in this first iteration you only do a shallow clone. Is that intentional?

I'm concerned that, if we don't have best-practice implementations in the engine and in Iota, we'll have a hard time fixing those up in the aftermath.

Also, we'll need to update all modules in Omega to not break the workspace. I think we wanted to throw out a couple of modules that are not working anymore, which will at least reduce the scope slightly...

[skaldarnar]

why is this change necessary?

[DarkWeird]

Because:

1. @DoNotAutoRegister is part of TS, not gestalt.
2. It fails with EmptyComponent. really abstract classes is not functional for ECS.

[skaldarnar]

Long line warning, add line break.

[DarkWeird]

Provide shallow and deep copy for collections.
shallow-copy for collections with immutable copies (String, primitives, enums, EntityRef,Assets,AssetData,Block... and custom immutable types)
deep-copy for several collections, where mutable types is using.