Exclude docs from crawling

Per mission from Dave, attached patch adds '-nodocs' to the buildsite
script to update the site without the docs. Since it's a lot faster, and
docs don't update all that often.

It also changes how the exclusions are done - now it does regexp match
anchored at the URL prefix. Previously, it would filter on any occurance
of for example "files" - with "docs" that turned into something *very*
likely to happen..

//Magnus