Re-evaluate Strategies Singleton vs. functions with xml_parsable: bool flag

[lu-pl]

The basic problem is that XML 1.0 needs an adapted text strategy because hypothesis.strategies.text() produces non-XML parsable strings.

lodkit.testing_tools.strategies is currently implemented using a Singleton approach defining a singleton using hypothesis.strategies.text() as a literal strategy and another singleton using an XML parsable text strategy for literals.

The singleton approach is kind of awkward because it introduces a class of properties for configuring a single parameter (xml_parsable: bool) which only affects literal strategies, namely triple_literals_plain, triple_literals_lang_tagged, triple_literals_xsd_typed.

Another option would be to have all lodkit strategies defined globally and move the xml_parsable: bool to the affected strategies, i.e. literal strategies and any strategy using a literal strategy...

Strategies requiring an xml_parsable_literals: bool flag would be:

• literal strategies

  • triple_literals_plain
  • triple_literals_lang_typed
  • triple_literals_xsd_typed
  • triple_literals

• triple strategies

  • triple_objects
  • triples

This is related to #10. Obviously, if testing support for XML serialization and parsing was dropped, this wouldn't be an issue.