feat: fix ExecutorCompletionService workaround in ChunkProcessingPipeline

[BenjaminAmos]

Contains

This pull request removes the work-around in ChunkProcessingPipeline where it reflects into the internals of the built-in ExecutorCompletionService class. This is no longer possible with newer Java versions and was never guarenteed to work in the first place.

Instead of making assumptions about ExecutorCompletionService, this pull request replaces the usage of the class with a new one specialised for dealing with chunks. Essentially, this makes sure that the submitted tasks are always of type PositionFuture, as several other classes in the codebase depend on the task future being an instance of that type. I believe that PositionFuture is used for certain optimisations related to the chunk load order.

This contributes to #3976.

How to test

• Run the game
• Create a new world
• Run the newly created world
• The world generation process should complete successfully
• Chunks should load around the player in-game

[skaldarnar]

Tested this out with Iota, world loaded quite the same as before this change, chunks appeared around the player, even after teleporting to a completely different location that was not generated before.