XML processor should produce HTML to enable website caching

[mikey]

Instead of just creating the XML document tree for the current request, the XML processor should also transform this into HTML, which means effectively it will be merged with the current XML post interceptor.

Such a refactoring has several benefits:

• website cache becomes applicable for the XML/XSL view engine resulting in a huge performance gain
• no incomplete XML documents from XML processor
• no need to wrap the response into a stubXMLResponse any more
• configuration issues about the need of adding the XML post interceptor will be solved

To achieve this and make this huge change manageable it will be split up into several smaller tickets:

• move skin generation from xml post interceptor into its own class
• move transformation of dom tree with skin from xml post interceptor to xml processor
• refactor xml processor to enable global hooks available on all pages

This is a prerequisite for enhancement #133.

[mikey]

Implemented.