Design separation of config and content parsing and model structure

Driver for HCM-37

Things to consider/decide/design:

• in how much content definitions are more restricted/limited than config definitions, e.g. not supporting merging, no persistence baseline, etc.
• how content definitions are applied to, and later extracted from, jcr, independent from config processing
• how content definitions are going to be 'packaged' in jars, e.g. combined with or separate from config definitions
• think of how to support 'force reload' (if deemed needed)