Port OreSpawn 4.0.16 to Forge 1.14.4

[SkyBlade1978]

Summary

Ports the accepted OreSpawn 4.0.16 campaign to Forge 1.14.4, ending at 4.0.16.114041.

Functional changes

• moves the single StoneReplacer pass to the start of LOCAL_MODIFICATIONS, immediately before provider surfaces, so provider-declared natural terrain is replaced before later authored features;
• rejects air, fluids, bedrock, and block-entity positions while preserving structures, vegetation, undeclared third-party blocks, and existing chunks;
• evaluates Stable Layers rock bounds against actual world Y while retaining shifted formation Y for layer/lithology identity and deterministic fallback selection;
• adds immutable Ore-dimension biome ID and dictionary include/exclude builder/getter contracts through the existing provider JSON fields;
• accepts canonical namespaced and legacy unnamespaced geome IDs in the editor;
• inspects top + 1 for exposed one-layer Snow while preserving buried/authored Snow and Ice;
• advances reports, documentation, Maven coordinates, workflows, and artifacts to 4.0.16.114041.

Forge 1.14 uses static, two-dimensional biome APIs and has no server GameTest harness, so the 4.0.11 dynamic-registry, 4.0.12 sampler-height, 4.0.15 3D biome-attribution, and GameTest-specific 4.0.16 repairs are explicitly recorded as target-inapplicable. Ordinary benchmark auto-stop remains unchanged.

Target preservation

• Minecraft 1.14.4 / Forge 28.2.26
• ForgeGradle 7.0.34 / Gradle 9.6.1
• MCP snapshot 20190719-1.14.3
• Java 8 bytecode and SRG reobfuscation
• pack format 4, API major 1, schemas 4/6/5
• existing provider meanings, profile filenames, migrations, registry IDs, chunks, and [4.0.6,5.0.0) compatibility
• native matrix-free UI, direct navigation, translated captions, duplicated Eclipse bin/main discovery, and the sealed Mineralogy 1.14.4 oracle

Validation

The exact candidate 7daa13b65b095f8e54da3f4504059101f91ae413 received manual acceptance and passed:

• 130 tests across 33 suites with zero failures, errors, or skips;
• 26 parsed JSON/mcmeta files, 16 locales with 357 matching keys, and byte parity for all 21 documentation exports;
• the complete offline build, check, Javadocs, release-artifact, checksum, Eclipse-run, and production-classpath gates;
• a cold exact-commit lifecycle plus empty-cache online Forge bootstrap and same-cache offline replay with zero Mavenizer source patches;
• exact reobfuscated Forge 28 packaged client and fresh/reload server runs, including world creation, editor round trips for all eight long fields, rendering, save/reopen, and clean shutdown;
• fresh/reload surface, natural-host, ceiling, fluid, unsafe-sentinel, Stable Layers, and Snow/Ice qualification;
• Mineralogy migration, OS4 profile precedence, byte-stable reports/reload, and preferred seed -4965128775892001975;
• staged three-run 81-chunk Sky/Average benchmarks with vanilla and Cyano controls; direct staged changes stayed below 5%, and the investigated normalized results remained below the 25% release blocker;
• independent Java-major, Forge/Minecraft metadata, pack-format, access-transformer, documentation/locale, and leakage audits.

The primary and cold builds produced byte-identical public jars. Main artifact SHA-256: F4BF7391A17730822AC6E0FAF7296A547CA850F2BC902E89C662EE19A8F0D911.

This PR does not tag, publish, or directly modify an MMD target branch outside the normal review/merge flow.