Sub-Biome Criteria Overhaul

[Apollounknowndev]

Especially after the addition of the any_of and all_of criterion, the sub-biome criteria system has been a bit of a mess. This PR aims to resolve this by moving criteria types into a registry-like system and simplifying the process of using sub-biomes in both code and json.

Internal Changes

The SubBiomeMatcher, SubBiomeMatcherImpl, and SubBiomeMatcherMarshaller classes were removed. The primary classes in their place are:

• Criterion: An interface for a criterion instance. It has three methods: getType for returning a CriterionType, getCodec for returning the criterion's codec, and matches for returning whether or not the criterion is matched. Anything that previously took a SubBiomeMatcher now takes a Criterion. Each criterion type now exists in its own class with its own codec in impl.
• CriterionType: An interface for a criterion type. Used to make criterion type-specific codecs work and not need one codec to have every field for every criterion type. It has two methods: getCodec for returning the criterion type's codec, and getId for returning the criterion type's identifier. Criterion types can be created using CriterionType.createType(codec, id).
• CriterionTypes: A class holding the map of identifiers to criteria types. Criterion types are registered here with CriterionTypes.add. The BiolithCriteria class in impl registers all built-in criterion types
• CriterionBuilder: A helper class with several constructed criteria (BEACHSIDE, RIVERSIDE) and all the methods needed to build criteria in code.

External Changes

Mods

Any references to the SubBiomeMatcher class itself should be changed to CriterionBuilder. Any references to the class in method references should be changed to Criterion. Examples of changes:

• SubBiomeMatcher.BEACHSIDE -> CriterionBuilder.BEACHSIDE
• SubBiomeMatcher.of(Criterion...) -> CriterionBuilder.allOf(Criteria...)
• BiomePlacement.addSubOverworld(RegistryKey, RegistryKey, SubBiomeMatcher) -> BiomePlacement.addSubOverworld(RegistryKey, RegistryKey, Criterion) On top of this, the inverted field in criterion has been removed. Invert criteria using CritereonBuilder.not.

Datapacks

• The criteria array is no longer wrapped in a matcher object.
• The criteria field can be a single criteria object instead of an array.
• All type fields need to have their ids prepended with biolith: (e.g. value -> biolith:value)
• Several misc. field renames to be more descriptive (exhaustive list coming at some point:tm:)

Todo

• Documentation in code
• Documentation in wiki
• More rigorous testing of changes (especially around ratio and alternate criterion types)

[gniftygnome]

This is timely, since I was going to stabilize the API as soon as I figure out the End issue I'm working on. Also, it's pretty much all stuff that's been on my to-do list for some time. From a very quick review, I think it can be merged as-is followed by a few minor code style changes. I need to:

• [x] Do a slightly more thorough review
• [x] Double-check ratio and alternate
• [x] Check all those hard-coded 64 values to make sure they're all applied to noise which cannot exceed them
• [ ] Update the wiki and the Biolith Examples project (non-blocker for this)

[gniftygnome]

I don't think Criteria was necessarily wrong. The problem is a Criterion can be a container of multiple criteria. When you talk about the object it probably makes more sense for it to be called Criterion, but when you are passing them around as arguments it usually makes more sense to call them criteria... I think this is actually why I created the Matcher as a container of criteria in the first place. Then somebody had to ask for sub-containers........ :laughing:

[gniftygnome]

I want to get this released soon and I have time today, so I'll just merge and adjust. But please feel free to ping me with any comments.