Update grammar to correctly produce value token

Problem was before that for example the input [0x000KKK] would produce a
hexadecimal and a value token (for the respective parts)

However, the more correct way is to have the whole input identified as a
value token. basicly if the whole input between "=" and a special
"end-of-assignment" token can't be identified as a string,bool,integer
whatever. it should be identified as a value token.

test/corpus/assignments.txt

@@ -10,11 +10,9 @@ KEY_WITH_VALUE = value
 
 (document
  (assignment
-  key: (identifier)
-  value: (value))
+  key: (identifier))
  (assignment
-  key: (identifier)
-  value: (value))
+  key: (identifier))
  (assignment
   key: (identifier)
   value: (value)))