[Intel-gfx] Fix for video spanning two heads

[Carl Worth]

With the video-tearing work I just did causing me to test multi-head
video I noticed that a video window spanning two heads gets rendered
incorrectly. Eric kindly pointed me at the relevant function,
(i830_clip_video_helper), and I found a one-line change there that
seems to do the trick.

Ian also kindly pointed me to bug 20980 which originally reported
the problem, (just a week ago). So the subsequent patch should
allow us to close that bug.

Eric, do you see anything cleaner we should do here? The only
real problem I have with this code is that i830_clip_video_helper
is an awful name---for the textured case, we're not doing any
clipping and "helper" is a decidedly non-descriptive verb.

But this seems to just be a mimic of the xf86XVClipVideoHelper
function in the X server with the same naming bugs. (Even without
clipping we seem to need to call this since "helper" apparently
also includes transforming integer quanitities into 16.16
fixed-point values).