BUGFIX: Skip content childnodes when returning document tree nodes

[Sebobo]

What I did

Speed up response time for return nodes for the document tree by a factor of up to 5x when uncollapsing a document or searching. With ~200 documents (e.g. Blog folder with many articles) the following results were achieved:

• Uncollapse blog page with ~200 articles:
  • Response size is reduced from 162 KB to 132 KB -> -19%
  • Response time is reduced from 2.5s to 0.6s -> -75%
• Search in the document tree for the same blog posts (also ~200 results)
  • Response size is reduced from 193 KB to 163 KB -> -16%
  • Response time is reduced from 2s to 0.5s -> -75%

How I did it

• Skip content child nodes if we only request document types depending on the clients requested usage
• Don't recalculate nodetype filter for each rendered node

How to verify it

• Tests still work
• Node trees behave the same way except that uncollapsing and search are much faster

Resolves: #1508

[Sebobo]

The follow up for this topic would be to get rid of all children in the response and only provide a parent id for building the tree and a hasChildNodes flag for the collapsible state check.