Docs sweep: README + KB catch up with the full session's work

Coverage audit of this session's ~30 commits against the docs found
gaps; fixed:
- README §Multiplayer was stale (only the old mesh flow, 'combat in
  progress'): now leads with the relay + operator-console path
  (internet play, join.bat/play_solo.bat, patient wait) and lists what
  the mesh path has verified (4-pod, camera seat, callsigns, clock);
  README controls line updated (turn/twist/pitch, pad, CONTROLS.MAP)
- decomp-reference env table: added BT_CAM_LOG, BT_SPEC_LOG (were in
  code, not the hub); BT_INPUT_LOG note gains the twist/elev axis
- pod-hardware §input remap: the AXIS SIGN CONVENTION finding
  (positive sim = CCW/left; bridge negates once per channel;
  screenshot-forensics lesson) was only in commit cb82d8c -- now durable
- multiplayer.md: PATIENT WALK-UP + MISSION CLOCK sections (features
  shipped in f9f230c/f57d25f, only incidentally referenced before)

checkctx CLEAN (20 topics, 68 glossary keys).  All session features
now have topic-file coverage.

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

context/decomp-reference.md

@@ -268,7 +268,9 @@ default-ON (`'0'` disables).
 | `BT_RELAY=<host:port\|auto>` | **D1 relay mode** — dial the relay/operator console instead of the P2P mesh; `auto` = LAN UDP-broadcast discovery on udp/15999 ([[multiplayer]] D1) |
 | `BT_SELF=<[pilots] tag>` | claim a specific roster seat in relay mode; absent/`auto` = the relay assigns one (SEAT_REQUEST) |
 | `BT_RELAY_TCP_ONLY=1` | disable the relay UDP channel (all unreliable traffic rides the TCP relay conn) |
-| `BT_INPUT_LOG` | binding-engine trace: CONTROLS.MAP load, button/keypad emissions, pad connect ([[pod-hardware]] §Desktop input remap) |
+| `BT_INPUT_LOG` | binding-engine trace: CONTROLS.MAP load, button/keypad emissions, pad connect, torso twist/elev axis ([[pod-hardware]] §Desktop input remap) |
+| `BT_CAM_LOG` | camera-seat/broadcast trace: streamed camera network count, director pick + Players-group census, ship follow state, PlayerLink dispatch/receive, ranking-window render ([[multiplayer]] camera seat) |
+| `BT_SPEC_LOG` | manual spec-audit dump at subsystem ctor: torso speed/limits, heat-sink count, reservoir capacity, per-subsystem coolant loop ([[pod-hardware]] §Spec cross-check) |
 
 Full render/locomotion gates (BT_RAMP, BT_MATPRI, BT_CULL, BT_SHADOW_*, BT_LODSEL, BT_ADDLOD,
 BT_PUNCH, …) are catalogued in [[rendering]]. Warp visuals: [[translocation-warp]].

context/multiplayer.md

@@ -326,6 +326,22 @@ phase). Plan: `~/.claude/plans/partitioned-snuggling-piglet.md`.
   SEAT_REQUEST branch (btconsole.py). Verified: 8/8 stub tests (distinct seats, FULL on
   exhaustion, claim clears reserve, duplicate-HELLO refused) + 2-node localhost e2e with NO
   BT_SELF on either pod → both seated, full ladder, launch, UDP flowing.
+- **PATIENT WALK-UP (2026-07-18) [T2]**: a pod dialed at a relay that isn't up yet no longer
+  aborts (release `Fail()` is a bare `abort()` = looked like a "crash", field report). Both the
+  seat request (`RelayRequestSeat`) and LAN discovery (`BT_RELAY=auto`) now RETRY (2s dials, 30
+  min cap) with live status into the join.bat console window (`BTRelayWaitStatus` attaches to the
+  parent cmd via `AttachConsole(ATTACH_PARENT_PROCESS)`): a banner + progress dots, a distinct
+  "game is FULL, waiting for a free seat" state, "seat assigned — joining!" on success. The pod
+  joins automatically the moment the operator starts the session; the unreachable MessageBox only
+  fires at the 30-min give-up. Removes the session-before-join ordering requirement. Verified:
+  pod launched at a dead relay, waited 20+s, relay started, pod seated + reached RunningMission
+  with no user action.
+- **MISSION CLOCK (2026-07-18) [T2]**: the egg `[mission] length=` (600 = the 10-min pod mission)
+  drives the pod countdown + the final-30s ranking window; the relay now SENDS
+  `Application::StopMissionMessage` (clientID 4, msgID 6, exitCode 0) at length+2s grace, or on
+  the operator's ⏹ END MISSION button (stdin `stop`). Was half-implemented — nothing fired it in
+  1995 the console owned the stop. Pod runs the authentic fade → mission-loop exit → clean
+  process exit. Verified live (40s test mission ended on the clock).
 - **OPERATOR CONSOLE (2026-07-18)**: `tools/btoperator.py` (PySide6) — the lost 1995 operator
   station recreated: mission editor with ALL values validated live from BTL4.RES
   (`tools/eggmodel.py`: maps=type14∩26, colors/badges/patches from the type-25 vehicletable,

context/pod-hardware.md

@@ -47,7 +47,16 @@ Axes: Throttle (lever, rate-walks + sticks), LeftPedal/RightPedal/Pedals (turn,
 JoystickX (torso twist), **JoystickY (torso ELEVATION — pitch aim; every 1995 control mode
 routes it to Torso::SetAnalogElevationAxis; keys R/F + pad LeftStickY in the default map;
 verified to the authentic VerticalLimitTop clamp, 20° on the Blackhawk; aims the GUNS — reads
-on the HUD elevation tape, the eyepoint stays level)**. Emissions mirror the RIO conventions above exactly; the four fire
+on the HUD elevation tape, the eyepoint stays level)**.
+**AXIS SIGN CONVENTION (2026-07-18, user-verified) [T1]:** the sim uses MATH convention —
+positive turnDemand / twist / free-aim slew is **CCW (LEFT)**. On the pod the RIO Ranger owned
+the hardware sign; the desktop key bridge negates ONCE per channel (`key_turn = -gBTDrive.turn`;
+`stickPosition.x = -gBTTwistAxis`; the SetFreeAimSlew call-sites pass the bridge value un-negated)
+so input-right = turn/twist RIGHT (manual p8: "pulling your joystick to the right torso twists
+your 'Mech to the right"). Forced/BT_GOTO harness demands are already sim-frame and stay
+un-negated. Lesson: large-rotation screenshot comparisons are AMBIGUOUS — trust tracked
+landmarks, numeric yaw telemetry with an established convention, or the user's live observation.
+Emissions mirror the RIO conventions above exactly; the four fire
 buttons (0x40/45/46/47) publish as levels to the bring-up fire channels instead of buttonGroup
 (avoids double-fire once the streamed ChooseButton mappings are exercised). XInput loads
 dynamically (xinput1_4 → 9_1_0; disconnected-pad probing rate-limited to 1 Hz). **Keys claimed by