Update CHANGELOG.md (#38170)

One column name updated to "New `max_token` for Generation" in CHANGLOG.md

sdk/evaluation/azure-ai-evaluation/CHANGELOG.md

@@ -69,7 +69,7 @@ outputs = asyncio.run(custom_simulator(
   - `RetrievalEvaluator`
 - The following evaluators will now have a new key in their result output including LLM reasoning behind the score. The new key will follow the pattern "<metric_name>_reason". The reasoning is the result of a more detailed prompt template being used to generate the LLM response. Note that this requires the maximum number of tokens used to run these evaluators to be increased.
 
-    | Evaluator | New Token Limit |
+    | Evaluator | New `max_token` for Generation |
     | --- | --- |
     | `CoherenceEvaluator` | 800 |
     | `RelevanceEvaluator` | 800 |