TURN RECORD: THE DISPLAY HALF STILL FAILS; TWO STRUCT FACTS CORRECTED.

A) QueryExport RETURNS PIXELS, NOT A PIXMAP
   xieQueryExportReply (XieProto.h) = { type, _reserved, sequenceNumber, length,
   pixel_count, lut_id, lut_mapping, _pad[15] }.
   So minor 24 reports how many colormap pixels the export allocated (plus an
   optional LUT id).  There is NO pixmap id to XCopyArea -- "export paints
   straight to the drawable".  My earlier plan of copying a pixmap to the window
   was based on a wrong assumption and is dropped.

B) XieExport.3X ERROR TABLE (read in full)
     BadIDChoice  Invalid identifier
     BadAlloc     Insufficient server resources
     BadDrawable  Invalid X11 drawable
     BadGC        Invalid graphics context
     BadMatch     source image is not constrained / gc not compatible with the
                  drawable / LUT has fewer levels than the source / LUT not
                  compatible with the drawable levels / source image not
                  compatible with the drawable
   Note there is no BadValue in this list, yet a correctly-idded export returns
   err2 (BadValue), so the failure is not one of the documented export
   conditions.

C) THE BadDrawable WAS MY BUG
   In the previous A/B run I hard-coded window/gc ids (0x55300/0x55400) instead
   of using c.new_id(), so the drawable id was never valid.  Re-run with real
   ids:
     white/black: transport=ok  puttile=ok  export=err2  src_base=00000000
                  trace RP/TYPE5
   i.e. BadDrawable disappears; BadValue remains.

D) BOUND PATH WITH THE CORRECTED OPERAND ORDER
   xieBindPhotomapReq = { photoflo_id, photomap_id }.  Sequence
     create photomap(w6=1) -> SetTransport(mode=3) -> PutTile(ok)
     -> create photoflo(type=1, w6=1) (ok) -> BindPhotomap {flo,map} (ok)
     -> Export -> ExecutePhotoflo(13)
   gives:  export = err1 (BadRequest) x6, execute = err1 x6, renderer trace
   (none).  So once bound, minor 21/13 are rejected outright -- the bound path is
   not usable with the operand shapes I have.

E) WINDOW IS PROVABLY DRAWABLE
   Control (unchanged): core-X PutImage into the same 64x64 window produces a
   visible block in dwt/screenshot.py.  So any successful XIE paint would be
   visible; the absence of change is real.
   White vs black captures continue to differ only by the Terminal Manager's own
   figures and the cursor (2697-2950 px).

STATUS.  Still NO client-chosen XIE image.  src_base (rp+0x610) 00000000 in every
run; the window is unchanged by XIE; xshow (core X) remains the only viewer that
has displayed an image.

WHERE IT STANDS: the import half is complete and verified by the server's own
QueryResource (comp_COUNT=1, 64x64, constrained).  The display half has exactly
two observable failure modes left -- unbound photomap => renderer entered but
src_base 0, bound photomap => BadRequest on export/execute -- and the documented
error conditions in XieExport.3X do not cover either.
