BT410: the MFD wave landing report (5.3.33-38)

Where the cockpit stands per head, the five defects the rig found and the
order they fell, and the measurement lesson: a large overlapping error makes
a small one measure as already-optimal, which is exactly what I recorded for
the title banner before the strip art was fixed.

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

restoration/source410/BT_L4/GAUGE-BLOCK.NOTES.md

@@ -778,3 +778,85 @@ Everything in this list came from the per-head instrument in one sitting;
 none of it was visible in the whole-frame score, which moved less than a
 point across the entire session even as Eng1/2/3 went to exact.  Score the
 heads, not the frame.
+
+================================================================================
+5.3.33-38 -- THE MFD WAVE: the weapon panels close to a few dozen pixels
+================================================================================
+
+Where the cockpit stands against the shipped 4.10 binary (planes.py):
+
+    head      shipped     ours   missing    extra
+    Eng1        17981    17981         0        0
+    Eng2        17981    17981         0        0
+    Eng3        17981    17981         0        0
+    Mfd1        41784    41651       140        7
+    Mfd2        40799    40716        97       14
+    Mfd3        32146    32128        31       13
+    Heat        21374    20971       483       80
+    Comm        15656    13557      2099        0
+    overlay      3283     3325        55       97
+    sec         35423    36137       325     1039     <- the 3-D scene (btl4vid)
+
+Whole frame: 98.2% pixel-identical / 97% coverage, from 89.7% / 92% at the
+start of the session.  Six of the ten heads are exact or within ~150px.
+
+FIVE DEFECTS, ALL FOUND BY THE RIG, IN THE ORDER THEY FELL:
+
+1. MissileLauncher::DefaultData carried Subsystem's shared-data tables where
+   every sibling weapon carries MechWeapon's -- so EVERY weapon attribute on
+   a launcher resolved to nothing and the gauge read the unbound zero cell.
+   The tell was a fire-ready disc the shipped STREAK 6 panel draws and ours
+   never did, while [proj-state] showed the launcher itself at state=Loaded
+   recoil=0 level=1 bin=24.  A healthy weapon behind a dead binding.  ~7.6K
+   pixels.  This one reaches well past the cockpit: PercentDone,
+   TriggerState, WeaponState, DistanceToTarget and RearFiring were all dead
+   on both SRM6 launchers.
+
+2. The strip art (image_names[auxScreenPlacement], qblh0..qblh6 -- the mech
+   diagram with this subsystem's segment lit) drew at the panel ORIGIN
+   instead of the quadrant's middle.  Solved rather than guessed: our extra
+   pixels clustered at panel x 0..70 against shipped's missing at x 140..230,
+   and a cross-correlation of the two sets peaked at dx=+144 dy=-54
+   independently on all three panels.  +144 is 0x90 exactly.  ~10.9K missing
+   and ~6.7K extra pixels.
+
+3. The ammo readout was inert (ammoBinLink protected, no accessor).
+   ProjectileWeapon now publishes GetAmmoCount() and the cluster copies it
+   into a live cell each Execute -- the pattern it already used for jammed
+   and reloadSeconds.  The STREAK 6 panel reads 0024 against shipped's 0024.
+
+4. CoolingLoopConnection::Update wrote a constant 0, so every panel's state
+   box read "OFF" where shipped shows the loop digit.  heat.hpp now publishes
+   GetCoolantAvailable() / ResolveCoolingMaster() / GetCondenserNumber() and
+   the connection follows the binary.  The box reads 4 / A against shipped's
+   4 / A, pixel for pixel.
+
+5. SegmentArc270 computed segmentSpan = 0.75 in its ctor and NOTHING read it
+   -- the engine base has no span member and the class declared no Execute,
+   so the dial lit all 20 segments where shipped lights 15.  A derived
+   Execute now scales the fraction, draws, and restores it (the connection
+   only rewrites currentValue on change, so scaling in place would compound
+   and walk the dial to zero).  Extra: 679/462/461 -> 7/14/13.
+
+--------------------------------------------------------------------------------
+THE MEASUREMENT LESSON, WHICH IS THE REUSABLE PART
+--------------------------------------------------------------------------------
+Mid-session I measured the title banner at y+0xb3 as the OPTIMUM and wrote in
+this file that it was at most 1-2px off: +3 and -3 both scored worse, and the
+correlation peaked at only 30%.  That reading was CONFOUNDED.  The strip-art
+error (defect 2) was ten times larger and lived in the same bands, so it
+dominated the metric and hid a clean one-row error underneath.  Once the
+strip art was right, the title fringe isolated at exactly 553/553
+missing/extra per head and +1 collapsed it to nothing.
+
+So: fix the LARGEST error on a head first, then re-measure the small ones.  A
+big overlapping defect makes a small one measure as noise -- or worse, as
+already-optimal, which is what I recorded.
+
+WHAT IS LEFT, LARGEST FIRST:
+  * sec (325 missing / 1039 extra) -- the 3-D scene head.  This is btl4vid,
+    the world renderer, and the next big climb; the gauge work cannot move it.
+  * Comm (2099 missing / 0 extra) -- content the shipped Comm page draws that
+    we never build at all.  Zero extra means nothing we draw there is wrong,
+    so this is a build-it wave, not a fix-it one.
+  * Heat (483 / 80) and the MFD remainders (~270 / ~34) -- fine detail.