Correct XMP URI serialisation in PDF/UA ext schema

[MatthiasValvekens]

URIs and text follow different conventions in the XMP data model. While many parsers tolerate URIs encoded as text, it's technically not allowed. Without this fix, iText will output PDF/A+UA files in which the pdfuaid extension declaration (more specifically, the pdfaSchema:namespaceURI property) violates ISO 16684-1:2011, 7.9.2.2:

Property and structure field elements that have normal (non-URI) simple, unqualified values may be replaced with attributes in the rdf:Description element. This also applies to the pseudo-structure for general qualifiers, including the rdf:value element. The element and attribute forms may be mixed.

Emphasis mine.

The code diff is only one line, but I also added two tests: one comparison test to check general XMP generation for PDF/A+UA files, and one more granular test to specifically check whether the URI is serialised using rdf:resource, and not as an XMP text value.

(@michaeldemey this is the thing we briefly discussed at the airport today)