fix: one neutral grey, rings at measured optical depth, no Jupiter pole graft

[alowpoly]

An audit after the Saturn cleanup found the same kind of invented values elsewhere; this fixes them.

• One neutral grey. Bodies with no measured colour carried four different greys (one warm, on every comet). All 485 now take #9a9a9a, from one shared constant, and a test keeps it that way.
• Rings at their measured optical depth. A ring band now states its published normal optical depth only; opacity is 1 − exp(−τ) and the colour is the neutral grey. Jupiter's rings (τ under 10⁻⁵) and Neptune's (0.003 and below) all but vanish, and Uranus keeps its ten denser rings as thin lines. Neptune's Adams arcs, whose positions are not published, are gone. Quaoar's rings drop from a display 0.4 to the published 0.004.
• Jupiter's poles. The Juno photographs grafted into the Hubble map are removed; the grey grid marks the caps Hubble never observed.

Limits: the narrow-ring minimum width stays, so Uranus's rings are wider than true; Jupiter still mounts an empty ring layer; lighting floors on bodies without a published law are unchanged.

Prepared assets

• Baked Jupiter, Uranus, Neptune, Chariklo, Haumea and Quaoar (prepare-object.mts).
• Published the bytes, with the Sun's world files and the Saturn moon dots.
• Committed the refreshed inventory.json files, and no baked files.

Clean: giant-layer, ring, observed-surface, context-colour and site page tests for the six bodies (233 tests), package lint, no-hashes check.