Create Horse Power - CE 1.2.6

Fixes

• Fixed NeoForge path output being applied twice after the pathEvaluated KubeJS hook. With KubeJS installed and no listener registered, a Great path could incorrectly turn a 2.0x RPM multiplier into 4.0x, and non-neutral stress multipliers could likewise be squared.
• Defined pathEvaluated speed/stress values as final absolute path multipliers. Scripts may still replace them with setSpeedMultiplier() / setStressMultiplier(), but the crank now assigns the returned final values instead of multiplying them again.
• Fixed NeoForge path effects being lost when KubeJS is not installed. The no-KubeJS firePathEvaluated fallback now preserves the evaluator's absolute speed/stress multipliers instead of replacing them with neutral 1.0x values.
• Preserved Forge behavior by keeping the common no-op script hook aligned with the evaluator's existing path multipliers.

Regression coverage

• Added discriminating evaluator coverage for LEGACY, WORST_BLOCK, and WEIGHTED_AVERAGE using a mixed custom path whose expected speed and stress outputs differ by mode.
• Added coverage proving disabling path stress scaling leaves RPM, validity, coverage accounting, and efficiency unchanged while neutralizing only the final path stress multiplier.
• Added coverage proving the stress toggle cannot bypass minimum path coverage.
• Added Forge 1.20.1 and NeoForge 1.21.1 crank GameTests for the production-shaped case: 16 base RPM / 32 base SU on a Great path keeps the 2.0x RPM path bonus while final stress remains exactly 32 SU.
• Added a NeoForge no-KubeJS regression verifying OptionalIntegrations.firePathEvaluated() preserves a 2.0x speed / 1.10x stress result unchanged.
• Retained the NeoForge KubeJS-present runtime regression so both optional-integration branches are covered.