[Performance]: Inefficiency of `ODataResource` and current ODataWriter architecture

[habbes]

I opened this issue to discuss fundamental problems with the current architecture of ODataWriter and its reliance on ODataResource objects. The concerns are mainly concerned with performance. I would like to open a forum to get thoughts and feedback from the wider community. I have created this issue for visibility but will move the discussion to this discussion forum: #2797

This is an experiment from the OData team to "design in the open" and get customer feedback early in the development cycle of major proposal, feature or change. I would like to get feedback from the community on this approach and your thoughts on what would be effective ways to gather community feedback during the design phase of major proposals.