updates

docs/pt2-lamp-selector-map.md

@@ -71,6 +71,69 @@ Follow-up `lamp-isolate-neighbor-single-boot` result:
 
 This confirms that the host/CCU can directly drive panel lamps through selector-table writes. It also validates using the ROM dispatch-neighbor list around `0x0007` and `0x0015` as a high-value lamp map.
 
+Follow-up `panel-atlas-standard-master-*` webcam runs:
+
+- `0x0012`, `0x0013`, and `0x0014` high-nibble/selected-bit sweeps did not
+  produce a clean STANDARD or MASTER lamp trigger. The `0x0013=0x8000` SLAVE
+  positive control still worked.
+- `0x0010`, `0x0011`, `0x0015`, `0x0016`, `0x0017`, `0x0018`, `0x0019`, and
+  `0x001A` high-nibble sweeps did not produce a clean STANDARD or MASTER lamp
+  trigger.
+- `0x0008` through `0x000F` high-nibble sweeps did not produce a clean
+  STANDARD or MASTER lamp trigger.
+- `0x0017=0x4000` lit the same far-right bottom BARS lamp/latch as the known
+  `0x0017=0x8000` family. Later `0x0018` rows in that run were latch-contaminated
+  and need fresh-boot isolation before assigning a separate meaning.
+
+Current implication: STANDARD and MASTER are probably not simple direct
+command-0 high-nibble writes in the `0x0008`-`0x001A` pocket, despite the older
+broad run that made MASTER flash. Treat that older observation as a real
+bench-visible event but not yet an isolated selector mapping.
+
+### `panel-atlas-big-visual-sweep-0001-017f-highbits`
+
+The broad webcam sweep and fresh-boot isolation pass produced these confirmed
+or near-confirmed panel-output mappings:
+
+| Selector/value | Current meaning |
+| --- | --- |
+| `0x0013 = 0x4000` | `IRIS/M.BLACK LINK` lamp |
+| `0x0024 = 0x8000` | LCD selector-button lamp |
+| `0x0024 = 0x0000` | LCD selector-button lamp remained visible at 0.5 s; not a simple clear in this timing window |
+| `0x0082 = 0x8000` | IRIS readout `OP` |
+| `0x0082 = 0x4000` | IRIS readout `1.4` |
+| `0x0082 = 0x0000` | IRIS readout blank |
+| `0x0083 = 0x8000` | MASTER GAIN `-3`, SHUTTER `OFF`, and IRIS AUTO |
+| `0x0083 = 0x0000` | same visible state remained at 0.5 s; likely latched/copied elsewhere |
+| `0x0093 = 0x8000` | BLACK/FLARE MANUAL plus white-balance PRESET |
+| `0x0093 = 0x4000` | BLACK/FLARE MANUAL plus white-balance AUTO |
+| `0x0093 = 0x2000` | BLACK/FLARE MANUAL plus white-balance MANUAL |
+| `0x0093 = 0x0000` | BLACK/FLARE MANUAL plus white-balance MANUAL |
+
+Run health was good: no resync events, no dropped bytes, and command-4
+readbacks appeared for the listed writes.
+
+ROM trace now confirms the `0x0013` bit split:
+
+- `E800[0x0013]` is `H'E826`.
+- Selector `0x0013` dispatches to `H'2E06`.
+- `H'E826.15` sets/clears `F791.6` and `F713.4`, matching SLAVE.
+- `H'E826.14` sets/clears `F791.5` and `F716.7`, matching
+  `IRIS/M.BLACK LINK`.
+- Local handler `H'200E` can also toggle `H'E826.14` from panel input
+  `F006.7/F6DB.7` when its session gate allows it.
+
+Interpretation:
+
+- `0x0082` is the cleanest direct readout lane found so far: set values update
+  the IRIS display and `0x0000` blanks it.
+- `0x0083` appears to be a combined status/display lane rather than only the
+  MASTER GAIN display. Its `0x8000` state also brings up SHUTTER `OFF` and IRIS
+  AUTO, and the clear write did not visually clear it at 0.5 s.
+- `0x0093` selects white-balance mode. In this run, `0x2000` and `0x0000` both
+  presented white-balance MANUAL, so bit 13 may be redundant in this context or
+  may need another gate to show a distinct state.
+
 ### `lamp-broad-status-selector-sweep`
 
 Visible result: