webbrowser speed

[Marco Martin]

hi all,
just as an experiment last days i've been playing with the web browser to see 
if some speed issues and interaction bugs can be fixed by changing how the 
rendering goes on.

(basically by letting webkit doing the scrolling instead of having a giant 
element with the whole page painted in it)

the result is mart/webbrowseractualscroll

is still pretty buggy, but it seems to already work way better and be 
basically in par with rekonq as speed, ie still not perfect, but way better 
than the current one, especially for large complex pages.

i'll now leave that branch as it is for a while, until we go for a web browser 
little sprint.

in the meantime if someone can give it a try to see if the findings are 
confirmed (ie faster than the current one and the mouse cursor doesn't get 
anymore "stick" to the page) would be great
(by building from sources on a desktop, definitely won't be packaged ;)

Cheers,
Marco Martin