- DHParser/parse.py: resume and skip directives now jump to the reentry-point right after the match of the regular expression for re-entry

- removed MetaParser._keep_node(). This kind of optimization is now done inplace which is considerably faster

- parse.py: OneOrMore, ZeroOrMore infinite loop detection replaced by a simple "break"; static analysis outcommented (never know when it might still become useful)

- DHParser/parse.py Parse.__init__() and Parse.reset() move clearing of cycle-detection buffer to __init__ to avoid infinite loop if reset() is called during the Parser.apply() for some reason

- DHParser/parse.py: bugfix MetaParser.return_values() empty nodes will now be properly deleted (unit test added)