DEC'S OWN IDS CLIENT SHOWS THE CORRECT DISPLAY SEQUENCE -- EXPORT INTO A PIXMAP.

Newly read (never used before): decimage-ids/ids_build_xie_pipe.c, DEC's own
client-side pipeline assembler.  The calls it makes:

  XieCrop, XieScale, XieRotate, XieTranslate   -- pipeline elements
  XieTapFlo(Photo_(xiectx), True)              -- TAP the photoflo (3 sites)
  XieSelectEvents(... XieM_ComputationEvent | XieM_DisplayEvent | ...)
  XieExport(Photo_(xiectx), CPix_(xiedat), RPixGC_(phoctx), 0,0, 0,0, w, h, ...)
  XieQueryExport
  XiePutImage
  XieFreeResource

Two findings, both new to this work:

1. **Export targets a PIXMAP, not the window.**  DEC exports into CPix_/RPix_
   (an X pixmap) and then the caller copies it.  This matches the manual's
   "XIEP_EXPORT creates a new server PIXMAP resource object" and explains why
   exporting straight to the window was never right.

2. **The photoflo is TAPPED (XieTapFlo) and elements are added to it**
   (Crop/Scale/Rotate/Translate) before export -- the pipeline model the manual
   describes, which the hand-built client never used.

IMPLEMENTED AND TESTED
  decclient now supports the pixmap path (argv[4]=1):
    XCreatePixmap(depth 1) -> XieExport(photomap, pixmap, gc, ...)
    -> XCopyArea(pixmap, window)
  Both fills run clean:
    exporting to PIXMAP / export sent (pixmap) / copyarea sent   (no X errors)
  but the framebuffer is still unchanged: white vs black AE = 2600 (TM figures
  and cursor only).

So the request sequence is now the real one, and it completes -- the pixmap is
simply empty because the accelerator performs no pixel transfer.

STATE
  * NO client-chosen XIE image has ever been displayed;
  * VRAM byte-identical for both fills (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.
