Update readme to include new env var section and update TOC.

Author: matherit

README.md

@@ -42,7 +42,9 @@ For notebook examples, see: [Amazon SageMaker Examples](https://github.com/awsla
 3. [Running the tests](#running-the-tests)
 4. [Pre/Post-Processing](#pre/post-processing)
 5. [Deploying a TensorFlow Serving Model](#deploying-a-tensorflow-serving-model)
-6. [Deploying to Multi-Model Endpoint](#deploying-to-multi-model-endpoint)
+6. [Enable Batching](#enabling-batching)
+7. [Other Configurable Environment Variables](#other-configurable-environment-variables)
+8. [Deploying to Multi-Model Endpoint](#deploying-to-multi-model-endpoint)
 
 ## Getting Started
 
@@ -612,6 +614,16 @@ SAGEMAKER_TFS_NUM_BATCH_THREADS="16"
 SAGEMAKER_TFS_MAX_ENQUEUED_BATCHES="10000"
 ```
 
+## Other Configurable Environment Variables
+The following environment variables can be set on a SageMaker Model or Transform Job if further configuration is required:
+
+```bash
+# Configures how long to wait in seconds for GUnicorn
+# to finish starting up before timing out.
+# Defaults to 30.
+SAGEMAKER_GUNICORN_SETUP_TIMEOUT_SECONDS="60"
+```
+
 ## Deploying to Multi-Model Endpoint
 
 SageMaker TensorFlow Serving container (version 1.5.0 and 2.1.0, CPU) now supports Multi-Model Endpoint. With this feature, you can deploy different models (not just different versions of a model) to a single endpoint.