Improve XPath predicate and filter evaluation

Consolidate code paths evaluating XPath predicates and filters.

Don't push context node on stack when evaluating predicates. I have no
idea why this was done. It seems completely useless and trying to pop
the context node from a corrupted stack has already caused security
issues.

Filter nodesets in-place and don't create node sets with NULL gaps which
allows to simplify merging a great deal. Simply move matched nodes
backward and create a compact node set.

Merge xmlXPathCompOpEvalPositionalPredicate into
xmlXPathCompOpEvalPredicate.

result/XPath/tests/chaptersbase

@@ -138,3 +138,31 @@ Set contains 0 nodes:
 Expression: //p[-100000000000000000000]
 Object is a Node Set :
 Set contains 0 nodes:
+
+========================
+Expression: //chapter[true()][position() mod 2 = 1][true()][2]
+Object is a Node Set :
+Set contains 1 nodes:
+1  ELEMENT chapter
+    ATTRIBUTE id
+      TEXT
+        content=chapter3
+
+========================
+Expression: //chapter[true()][2][true()]
+Object is a Node Set :
+Set contains 1 nodes:
+1  ELEMENT chapter
+    ATTRIBUTE id
+      TEXT
+        content=chapter2
+
+========================
+Expression: //node()[false()]
+Object is a Node Set :
+Set contains 0 nodes:
+
+========================
+Expression: (//node())[false()]
+Object is a Node Set :
+Set contains 0 nodes:

test/XPath/tests/chaptersbase

@@ -11,3 +11,7 @@
 //p[0 div 0]
 //p[100000000000000000000]
 //p[-100000000000000000000]
+//chapter[true()][position() mod 2 = 1][true()][2]
+//chapter[true()][2][true()]
+//node()[false()]
+(//node())[false()]