Consider rejecting unknown properties/attributes during parse

[Chris Fuller]

The library was intentionally designed to tolerate/ignore any properties or attributes that it does not support. The intent was to make it less likely to fail as new features are introduced in ADF. However, this now seems to be causing more problems that it is solving:

1. It leads to an inconsistent experience, as valid new properties/attributes get dropped without any warning.
2. It causes consistency problems because the content service will reject ADF content with unrecognized properties or attributes in most cases; when the library views such content as valid, the later rejection by the CS causes a failure much later in the conversion process, which is not where we want it.
3. It prompts us to use the schema validator to catch them instead, which is much slower and gives much less informative error descriptions.

Consider adding an allow list for both areas that the parser can use to reject the same things that the CS would.

‌