Performance improvements

[dpvc]

I've made a performance branch of my fork that includes two updates: the reduced processUpdateTime for better responsiveness, and the separate class for during the typesetting that IE can use to set position:relative to get measurements right.

See dpvc@f86e817
and dpvc@1e4efb3 for details.

Davide

[dpvc]

This is a duplicate of issue #17.

[shogun70]

Regarding the separate class for typesetting, the generated CSS rule:
.MathJax .MathJax_Processing span { position: relative; }
will never apply to anything, as the both those classes are on the one element.
Consequently this change makes the default position style to be "static" during and after processing.
Since "static" is the default anyway, we could remove the rule completely (assuming some diligent testing doesn't reveal rendering errors).

[dpvc]

OK, I've done more testing and I still can't find what the width problem was that the relative position was supposed to fix. It could be that a later change took care of it in another way (I'm afraid I can't keep track of all the IE bugs I've had to work around).

I removed the processing class. Anything else need changing?

[shogun70]

It would be good to have a test page where processUpdateTime is configurable, but I can't see any reason these shouldn't be pushed up now. So I'll do that ASAP.

[dpvc]

OK, I think I have merged this in. (I'm still a little awkward with git, so I'm not sure I did it in the most efficient way, but the changes are now there.) So I'm closing the issue.