feat(render): Phase A8 — indoor visibility + streaming fixes batch

Lands the working A8 indoor-rendering and streaming fixes accumulated this
session. User has verified these visually to some degree (e.g. lifestone /
translucent meshes confirmed fine under the FrontFace flip; bridge / wall /
collision regressions confirmed fixed after travel); not every path has been
exhaustively gated. The cellar-flap defect remains OPEN and will be solved
the retail-faithful way via a dedicated brainstorm (see handoff docs).

Rendering core (reviewed, high confidence):
- EnvCellRenderer SSBO stride fix: upload packed Matrix4x4[] (64B) instead of
  the 80B CPU InstanceData struct the shader never expected — fixes the
  transform/texture "explosion" for any draw with >1 instance (cells that
  dedupe to a shared cellGeomId). Real root cause.
- WB-style global FrontFace(CW) + per-batch CullMode carried through the MDI
  layout (GroupKey + BuildIndirectArrays + DrawIndirectRange split into
  same-cull runs with absolute uDrawIDOffset per run).
- EntitySet partitioning (IndoorPass / OutdoorScenery / LiveDynamic) +
  WorldEntity.BuildingShellAnchorCellId so building shells scope to their
  dat-derived building cell instead of rendering everywhere.
- RenderOutsideInAcdream (look into buildings from outside) +
  CollectVisiblePortalBuildings frustum cull of portal bounds.
- Sky-when-inside-building + per-cell audit probe + GL-state probe.

Streaming / perf (test-covered; not independently code-reviewed this session):
- Near/far priority queues so near work wins over far; PromoteToNear carries
  full landblock + mesh data; LandblockEntriesWithoutAnimatedIndex avoids
  rebuilding the animated-lookup dict in the hot draw path. Fixes the
  bridge-not-appearing / missing-walls / broken-collision-after-travel
  regressions and improves post-transition FPS.

Tooling + docs:
- tools/A8CellAudit: offline dat cell/portal/building dumper (portals +
  buildings modes) — reproduces the cellar-flap investigation with no launch.
- docs/research cellar-flap root-cause + option-2 handoff (the didInsideStencil
  double-duty finding + the WB-recursive design decision + brainstorm prompt),
  entity-taxonomy, replan, issue-78 visibility investigation.

Diagnostics retained on purpose: ACDREAM_A8_DIAG_* gates, portal_stencil.vert
provisional pos.w clamp, and the probe families are kept (env-var gated, zero
cost when off) because the pending option-2 cellar-flap brainstorm needs them.
Strip in the option-2 ship commit.

Indoor branch stays behind ACDREAM_A8_INDOOR_BRANCH=1 (default off = pre-A8
visual). Build green; App tests + Core (streaming/dispatcher/loader) tests pass.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

docs/ISSUES.md

@@ -1192,7 +1192,7 @@ or +small fix if different. Not blocking M1.
 ## #71 — WorldPicker Stage B — polygon refine for retail-accurate clicks
 
 **Status:** OPEN
-**Severity:** LOW (Stage A — screen-rect picker — is sufficient for M1)
+**Severity:** MEDIUM (Stage A now causes real play mis-picks through open doors/windows)
 **Filed:** 2026-05-16
 **Component:** selection / picker
 
@@ -1212,6 +1212,15 @@ to the visible mesh — under-pick what looks like empty space inside
 the rect, catch visible mesh that pokes past the sphere boundary
 (creature outstretched arm, sign edge).
 
+**New evidence (2026-05-28 / Phase A8 visual gate):** User stood outside
+a Holtburg building, saw a vendor through an open doorway/window, clicked
+the visible vendor, and acdream selected the door instead:
+`[B.4b] pick guid=0x7A9B4015 name=Door`. This is exactly the Stage A
+failure mode: the open door's projected `Setup.SelectionSphere` rect is
+closer than the vendor's rect, even though the visible door polygon is not
+under the cursor. The fix is polygon refinement against visible GfxObj
+triangles plus current animated part transforms; do not special-case doors.
+
 **Acceptance:** Pipe per-part GfxObj visual polygons through a
 `PickPolygonProvider` interface (don't duplicate mesh decoding —
 hook the existing `ObjectMeshManager` cached data). Two-tier in
@@ -1222,8 +1231,8 @@ frame edges.
 
 **Estimated scope:** Medium (~4-6 hours). Defer until visual
 verification surfaces a Stage A miss in real play. The user
-confirmed 2026-05-16 that "I can click on longer ranges now so
-good" — Stage A is enough for M1's "click an NPC" demo.
+confirmed 2026-05-28 that the door/vendor case is now observable in real
+play, so this should be scheduled soon after A8 rather than left as polish.
 
 ---