docs(simulation): update step_canvas row-space doc — T-1186 is fixed (PR #206 review)

Hoshe's review finding: the build_step_canvas row-space convention block
still described region_centre_latitude_deg as pole-anchored/bugged and
said 'T-1186 stays open' — both false after this PR's own first commit.
Rewritten to reflect the fixed state; the deliberate signed-convention
rationale and the layer_proxy deferral (still true, still T-1181's scope)
are preserved.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

server/src/atlas/step_canvas.rs

@@ -896,34 +896,29 @@ fn resolve_canvas_extent(
 /// origin-anchored — see the row-space convention note below); required for
 /// every fixed rung (the canvas centre in world metres).
 ///
-/// **Row-space convention (T-1186 interaction — MUST READ):** the Global
-/// (rung-0) canvas uses **equator-anchored SIGNED region rows**
+/// **Row-space convention (T-1186 — resolved):** the Global (rung-0) canvas
+/// uses **equator-anchored SIGNED region rows**
 /// (`wy = (row - rows/2) * REGION_M`, negative = north), matching the canonical
 /// `derive_orbital_at_metres`/`derive_at_metres_with_riparian` convention
 /// (D-256(a): the ONE absolute-metre derive core; that core's own internal
 /// `district_pos` floor-division and `lat_frac` computation are both
-/// equator-anchored-signed). This is a DELIBERATE DEPARTURE from two
-/// existing non-negative pole-anchored row conventions already in this
-/// codebase: (1) `region_profile::region_centre_latitude_deg` (row 0 = north
-/// pole, `lat_frac` clamped to `[0, 1]`) — the exact function T-1186
-/// documents as bugged when fed the derive core's signed rows; (2)
+/// equator-anchored-signed). When this canvas was built, that was a
+/// DELIBERATE DEPARTURE from two non-negative pole-anchored row conventions
+/// then in the codebase — chosen because signed-equator rows are the only
+/// shape consistent with D-256(a)'s "one derive core, one inverse mapping"
+/// principle, and using either pole-anchored convention here would have
+/// reproduced T-1186's bug class in a third place. Since then: (1)
+/// `region_profile::region_centre_latitude_deg` — T-1186's actual bug, the
+/// baseline-latitude function this derive core reaches transitively via
+/// `region_baseline_at_district` — is **fixed** (PR #206) to the same
+/// equator-anchored signed inverse mapping, so the baseline latitude every
+/// canvas cell receives now agrees with these rows on every rung. (2)
 /// `layer_proxy::build_region_grid`/`LayerRegionOutput` (non-negative `ry`
-/// rows, `0..=max_region.1`) — the collapsed whole-body region layer D-256(f)
-/// explicitly deferred to THIS rebuild ("the overlay stays visibly stale
-/// until T-1181 replaces it — accepted, noted"). Signed-equator rows are the
-/// only choice consistent with D-256(a)'s "one absolute-metre derive core,
-/// one inverse mapping" principle — using either pre-existing non-negative
-/// convention here would reproduce T-1186's bug class in a THIRD place. This
-/// derive core does **not** fix T-1186 itself (the bug lives in
-/// `region_centre_latitude_deg`, called transitively via
-/// `region_baseline_at_district` inside `derive_orbital_at_metres`) — every
-/// Global-canvas cell at a genuinely northern (negative signed-row) position
-/// still receives T-1186's wrong (+90°, polar) baseline latitude today, same
-/// as the Region/District/Quarter/Block/Chunk rungs already do via the same
-/// shared derive core. T-1186 stays open, tracked separately, exactly as the
-/// ticket's KNOWN INTERACTION note requires — fixing it here would touch
-/// `region_centre_latitude_deg`'s shared convention, which is explicitly out
-/// of this ticket's scope (T-1186 is "a separate ruled convention change").
+/// rows, `0..=max_region.1`) remains the one pole-anchored outlier — the
+/// collapsed whole-body region layer D-256(f) explicitly deferred to THIS
+/// rebuild ("the overlay stays visibly stale until T-1181 replaces it —
+/// accepted, noted"); unifying it onto the signed convention is proposed
+/// for T-1181's scope, at which point one convention remains codebase-wide.
 #[allow(clippy::too_many_arguments)]
 pub fn build_step_canvas(
     seed: SeedChain,