Capture more of the Document

[dubinsky]

• handle Locator2, DTDHandler, DeclHandler, and LexicalHandler DTD and entity methods;
• introduce DtdBuilder and delegate DTD methods to it;
• introduce ElementContentModel and delegate parsing of the element content model in the DTD element declarations to it;
• XMLLoader.loadDocument() returns baseURI, XML version, encoding and standAlone flag, and DTD;
• tests demonstrating the new returns.

Now that XMLoader returns Document, there is a place to return more information abut the parsed XML document - and this change does that, bringing XMLLoader to parity with the ConstructingParser ; returned now are:

• document's baseURI (which ConstructingParser never returned and indeed couldn't return);
• XML version of the document (which ConstructingParser returns correctly only if it is "1.0");
• standalone status of the document;
• document's encoding (ConstructingParser seems unable to parse documents encoded in UTF-16);
• document's DTD (ConstructingParser does not return element and attribute declarations, notations and some entities).

To preserve binary compatibility, parsed DTD is converted (with some loss of fidelity) into pre-existing ContentModel classes, instead of replacing them with the newly-introduced ElementContentModel.
To avoid capturing by the binary compatibility rules, newly introduced classes are package-private.

[SethTisue]

no objection here — feel free to merge if no other reviewers appear

[dubinsky]

feel free to merge if no other reviewers appear

Thank you!