Bug 687665

The source PDF file contains an image.
In the resulting PS file, the image is not converted to
a PostScript "colorimage" as expected, but painted in a
very strange, and inefficient way (see below). The resulting
PS file is extremely huge. It looks like in the PS file,
each image pixel is drawn separately with the "fill"
operator, is it?

...
%%Page: 1 1
%%BeginPageSetup
GS_pswrite_2_0_1001 begin
511 383 null setpagesize
/pagesave save store 197 dict begin
0.1 0.1 scale
%%EndPageSetup
gsave mark
100 145 212 rG
310 3829 5 1 rf
102 146 212 rG
315 3829 9 1 rf
103 146 214 rG
324 3829 4 1 rf
106 145 212 rG
328 3829 9 1 rf
107 146 213 rG
337 3829 9 1 rf
108 149 214 rG
346 3829 10 1 rf
107 148 214 rG
356 3829 9 1 rf
103 149 216 rG
365 3829 4 1 rf
104 147 215 rG
369 3829 5 1 rf
105 146 215 rG
374 3829 2 1 rf
106 145 215 rG
376 3829 2 1 rf
108 146 217 rG
378 3829 5 1 rf
108 147 215 rG
383 3829 2 1 rf
109 148 216 rG
385 3829 9 1 rf
108 147 215 rG
394 3829 3 1 rf
107 144 214 rG
397 3829 2 1 rf
...