TURN RECORD: SCAN OF THE RENDER DRIVER AFTER 0x1B296.

Static scan of 0x1B296..0x1B6E0 for everything that could yield the A8=2 the
export handler returns:

  0x1B360  MOVK 0x2,A11 ; 0x1B362 MOVB *A11,*A5(0x58)
           -- this is the ELEMENT-TYPE store (type 2), NOT an error.  It writes
              the element type into the render-params block; I had earlier
              chased this as a possible BadValue and it is not one.

  0x1B3C4..0x1B43E  the render worker's helper calls
           CALLA 0x07FC229C (x5), 0x07FC1EB2 (x2), 0x07FC1F5E (x2), 0x07FC2210 (x2)
           -- the actual per-element processing; these return codes are the
              remaining candidates for A8=2.

  0x1B5FA..0x1B67A  drawable-bind path: returns MOVK 0x18 and then MOVK 0x9
           (BadDrawable), NOT 2.
  0x1B68C..0x1B6E2  gc-bind path: returns MOVK 0xd (BadGC), NOT 2.
  0x1B612 / 0x1B6A0 CALLA 0x07F174A4  (bind helper)
  0x1B672            CALLA 0x07EFFA5E  (send-error helper)

So there is NO `MOVK 0x2` that sets A8 anywhere in the scanned range: the A8=2
observed at AFTER_RD (0x6309E) is a value RETURNED by one of the 0x7FC2xxx
helpers and propagated out of the driver unchanged.

MEASURED (unchanged, reproducible):
  import: element+0x28 -> 0xFF (white) / 0x00 (black), QueryResource comp_COUNT=1
  export: all handler checks pass, context alloc OK (ret=00321C50),
          render driver called at 0x63098, returns a8=2, handler returns it
  window: 64x64, depth 1, provably drawable (core-X PutImage control)
  framebuffer: white vs black byte-identical (AE = 0)

STATUS.  NO client-chosen XIE image.  xshow (core X) remains the only viewer
that has displayed an image.

WHERE IT IS NOW: the last unidentified return is inside the 0x7FC2xxx helper
family invoked between 0x1B3C4 and 0x1B43E.  That is the only place left that
can produce the 2.
