Planet weather depth: atmosphere gate, acid rain, real weather control

[StannisMod]

Summary

Closes logical gaps in how planet weather interacts with vanilla/AR mechanics — no brand-new weather mechanics (no dust/sand storms), only compatibility fixes plus acid rain.

• Atmosphere gate. Planet rain/thunder is now gated on a configurable minAtmosphereDensityForRain (default 75): thin/airless worlds stay clear regardless of weather markers. Fixes the inconsistency where rainMarker=1 could "rain" on a vacuum moon.
• Thunder requires rain. updateWeather enforces that thunder can't exist without rain (vanilla couples them — lightning only strikes where it's raining, thunder strength is scaled by rain strength). thunderMarker=1 + rainMarker=-1 no longer leaves a dead storm flag.
• Weather Controller actually controls weather. The satellite/remote rain/dry modes now drive the planet's rain marker (rain → always, dry → clear) in addition to their existing water-flooding behaviour; flood leaves weather untouched. Marker is handed back when the controller is removed.
• Acid rain. New per-planet acidicRain flag (XML + NBT + client sync via PacketDimInfo). Unprotected players under open acidic rain take damage (acidRainDamage / acidRainDamageInterval config); a full protective suit shields them, independent of breathability. Falling rain is tinted green client-side via an EntityRenderer.renderRainSnow mixin.
• Surfacing. New /ar planet weather [dim] prints the weather profile, atmosphere-gate status, current state and the tick countdown to the next change. (Profile values are also editable through the existing /ar planet get|set.)
• Bug fix (ledger Many bugfixes, so many that I can not even list them all + terraforming update #1 #8). canDoRainSnowIce compared the /100-normalized density against the literal 75, so it was effectively always false — AR planets never accumulated rain/snow/ice through that gate. Now uses the raw 0..100 density vs the config threshold.

Note on vanilla weather

Verified against decompiled MC: vanilla weather is a single global flag, filtered per-biome at the manifestation level (snow vs rain vs nothing) — it is not async-by-biome, so nothing was re-invented there. Vanilla has no "dry storm", so that idea was dropped in favour of the thunder-requires-rain invariant.

Test plan

• testUnit + testIntegration green — added acidicRain NBT round-trip (DimensionPropertiesTest) and XML parse/default (XMLPlanetLoaderTest).
• testServer (filtered) 12/12 green — new PlanetWeatherGateTest (thin atmosphere suppresses rain despite rainMarker=1; thick allows it; thunderMarker=1+rainMarker=-1 → no thunder) + no regression in existing weather/atmosphere suites.
• Real client on display :100 (Mesa/llvmpipe GL): MixinAcidRainRender applies into EntityRenderer at runtime with no injection error; client boots clean.
• Not yet eyeballed: the literal green rain pixels on an acidic planet need a manual look in a live GPU session (visual-regression is out of scope). Risk is low — the tint is an identity multiply for non-acid worlds and the redirect resolved at runtime.

Compatibility

• The render mixin uses require = 0 so an OptiFine-rewritten renderRainSnow cannot abort the (required) mixin config — it silently skips the tint instead.
• No registry-name or NBT-shape breaks; acidicRain is additive and defaults to false.