feat(simulation): district moisture spatial gradient (T-1080)

moisture_q was a single body constant (derive_moisture_q took only &BodyParams →
80 for every one of 2048 ocean-world districts), so vegetation/terrain/ecotones
were uniform — "nothing to fuzz" (the T-1080 believability bug).

- district_profile.rs: derive_moisture_q is now a body ceiling (hydrosphere+atmosphere)
  × per-district gradient — latitude (equator wet → pole dry), elevation (orographic/
  rain-shadow), continentality (100 − ocean_fraction_q). Integer/D-010, no new plumbing
  (elev_q/ocean_fraction_q/latitude_deg already reach build_district_profile). Mirrors
  D-240's temperature model. Coefficients in ClimateConstants + climate_constants.toml
  [moisture_gradient] (provisional; Q-123 calibrates). Tuned so a wet body stays mostly
  green with drier patches and a frozen body clamps mostly barren. + gradient unit test.
- region_profile.rs: region-tier moisture gets the region-latitude gradient (elevation/
  continentality are district-scale).
- Propagates to precipitation_class / vegetation_class / morphology_zone (all read
  moisture_q). D-239 §2 amended. Believability golden regenerated.

Probe (Arbour @ yolo): moisture_q distinct 1 → 49 (spread 0 → 48); 5/7 → 6/7 criteria.
Edict (frozen): moisture 0..20 ×21 — gradient within its low ceiling. Remaining fail
(vegetation-present) is the metric dividing by all-sampled incl. water on a 2/3-ocean
world — a Q-123 calibration concern, not a climate defect.

clippy --all-targets -D warnings clean; 1580 lib tests + harnesses pass.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

server/data/climate_constants.toml

@@ -84,3 +84,16 @@ ice               = 0.85  # frozen surface — slight moderation
 subsurface_liquid = 0.9
 subsurface        = 0.95  # buried water — minimal surface effect
 subsurface_ice    = 0.95
+
+# ---------------------------------------------------------------------------
+[moisture_gradient]
+# Per-district moisture spatial gradient (T-1080, D-239 §2). Integer points
+# subtracted from the body's hydrosphere moisture *ceiling* so the field varies
+# across the body instead of being a single constant (the bug: moisture_q = 80 for
+# every district -> uniform vegetation/terrain). Wet at coast/lowland/tropics, drier
+# toward interior/highland/poles. Provisional magnitudes — Q-123 calibrates. Mirrors
+# the embedded ClimateConstants::default(); not yet loaded at runtime (TOML wiring is
+# T-1032), kept in sync by hand like the other tables.
+lat_penalty         = 20   # lost equator->pole  (x |latitude|/90)
+elev_penalty        = 15   # lost low->high elevation (x elev_q/100, rain-shadow)
+continental_penalty = 20   # lost coast->interior (x (100 - ocean_fraction_q)/100)