…g negative lookahead regex

…being, write to disk or write to stdout)

…tNode

…nd the NodeAttributes as a single object instead of passing individual pieces. Will make it easier to add more pieces to the NodeAttributes

…- if an Attribute is negated, accept the node not having any value at all

…n the nodes

…as lemma) from a node

…notationLookup keys

…s which are Maps.

Currently the syntax is @, subject to change

eg,

morphofeatures@foo=bar

In this expression, bar can be a regex, but foo and morphofeatures cannot.
It might be worth adding regex capabilities for both of those,
Also, !@ would be a useful addition.

This checks at Semgrex compile time (not Java compile time) that
the annotation used for key/value is actually a Map

Has a test that the error checking and a simple search both work.

Need to negate the = in the NodePattern output

This map allow repeated elements inside the same brackets

…and Ssurgeon

…. Actually, there may not even be a reason to disallow two positive constraints if they are both regex

…t is pointed out that it may be possible for a properly constructed deserializer to execute arbitrary code. See https://github.com/stanfordnlp/CoreNLP/security/advisories/GHSA-wv35-hv9v-526p

Check for HasLemma when converting to make a possible error a bit more readable