STAGING-BUFFER WRITERS, BY PC, WHITE vs BLACK.

Watched writes to 0x004FE780..0x004FEB7F with the writing PC, one export per fill:

  white : 1588 writes   PCs 3F229B80 x724, 3FB44CC0 x224, 3F70C9A0 x162
                        values 0000 x852, 5555 x64, 4100 x55
  black :  150 writes   PCs 3F229B80 x106, 3F70C9A0 x16, 3F70D580 x16
                        values 0000 x82, 4100 x6, 40E0 x4

The two runs differ in volume and in writer mix, but neither carries the tile:
no 0xFFFF appears in either, and the dominant writer (3F229B80, file 0x114D70) is
present in both.  The traffic is allocator/manager bookkeeping, not pixels.

Combined with the earlier measurements this closes the question of where the data
goes: the fill-dependent path ends at the CPU read of the engine output
(0x5600200: FFFF x256 white / 0000 black).  Nothing after it -- staging buffer,
VRAM -- is fill-dependent, and the copy that would carry it is not something I
can identify from the observable firmware.

STATE (unchanged)
  * NO client-chosen XIE image has ever been displayed;
  * VRAM byte-identical for white and black exports (0 differing bytes / 128 KiB);
  * model opt-in (VT1200_ACCEL=1); default configuration unchanged;
  * xshow (core X) is the only viewer that has shown an image.
