[Pixman] [PATCH pixman] V7 implement PIXMAN_FILTER_GOOD/BEST

[spitzak at gmail.com]

Differences from previous version:

- Added compile-time option to pipe filters to gnuplot for analysis.

- Fixed integral when one of the filters is IMPULSE, previous version was
offsetting the sample from the center.

- Distribute fixed-point error on all samples, rather than putting it all
on the center.

- If reconstruct=impulse and scale < 1 (or sample=impulse) then it was
producing garbage by attempting to renormalize filters that should not
sum to 1. Even the "correct" result is useless, so I just made this set
the scale to 1.

- Comments and formatting fixes.

I consider all patches except the last one to be bug fixes. The last
patch changes the output for GOOD or BEST filter settings. Output is
nearly identical towhat Cairo is producing now, but has some minor
tweaks to cutoffs between filters and handling of very small scales.
This also detects more cases where nearest or bilinear can be used as
a fast path.

With all of these applied, the filter generation can be removed from cairo as
pixman will do it. The fallback to image for the X11 backend could be removed
if it is known the X11 server is using this new version.

This is also a necessary step in order to introduce 2-pass image filtering
to pixman. Currently software such as Cairo cannot use the filtering in
pixman and this means it will bypass any such improvements.