[nichinichisou0609]

Problem

The chart data page currently consumes sampled chart-stats/2 fields. New chart-stats/3 documents publish nominal expectations and numerical enclosures, so reading them as seed samples loses chart figures and mislabels their uncertainty.

Changes

• Read chart-stats/3 expectation fields and keep replay seed selection independent from statistical figures.
• Preserve measured rank-1 bonus expectations. Change only the selected ranges when applying other ranks, without truncating a mean as though it were an individual native score.
• Validate used interval values and dimensions, reject unsupported nonlinear rank extrapolation, and reject malformed new-format records without falling back to stale seed fields.
• Expose raw numerical enclosure radii separately from legacy sampling SE. Scenario transformations do not advertise an unmeasured enclosure.
• Update chart details, aptitude displays, Chinese/English explanations and the guide. Existing sampled files retain their original interpretation while deployments transition.

Validation

• 53 passing tests across songs expectations, songs page, Pareto and replay presets; one existing optional WASM test requires a separately configured engine.
• Eleven new expectation cases, including a fixture containing unmodified native chart-stats/3 values from a real mixed Just/Luck/Combo chart.
• node --check examples/songs/songs.js and git diff --check
• The coordinated nnnotes publication smoke gate passes against these modules, including independent nominal aptitude-check reconstruction.

The coordinated nnnotes.music-data/2 producer in StarMoe-org/nnnotes#21 pins this PR's exact source commit for its page smoke gates. Native program caching is in empty-sekai/ournotes-deck#44, and the site consumer is nichinichisou0609/moenotes#7. Deploy compatible readers before publishing the new data format.