Add tip in the docs on ambiguity investigation

docs/disambiguation.md

@@ -440,3 +440,9 @@ with pytest.raises(ParseError):
     `custom_token_recognition` can be used to implement custom lexical
     disambiguation by calling `get_tokens` and then reducing returned list to a
     list with a single result.
+
+
+!!! tip
+
+    See the end of [the parse trees section](parse_trees.md) for a tip on how to
+    investigate ambiguities in GLR parsing.

docs/parse_trees.md

@@ -71,4 +71,20 @@ with `tree_str()`:
 
     You can use `tree_str()` on the root of the parse tree to get the string
     representation of the parse tree. This can be handy to compare multiple
-    trees returned by GLR parser to analyse ambiguity.
+    trees returned by GLR parser to analyze ambiguity.
+
+    For example:
+
+        # Here we have GLR parser which returns multiple parse trees due to ambiguity.
+        # It is important to turn on tree building.
+        parser = GLRParser(g, build_tree=True)
+        trees = parser.parse_file('some_file')
+        for idx, tree in enumerate(trees):
+            with open(f'tree_{idx:03d}.txt', 'w') as f:
+                f.write(tree.tree_str())
+
+    Now you can run any diff tool on the produced outputs to see where are the ambiguities:
+
+        $ meld tree_000.txt tree_001.txt
+
+