Minor changes attempting to fix Spark unit tests

Author: sapessi

aws-serverless-java-container-core/src/main/java/com/amazonaws/serverless/proxy/internal/servlet/AwsLambdaServletContainerHandler.java

@@ -166,7 +166,7 @@ protected void setServletContext(final ServletContext context) {
         servletContext = context;
         // We assume custom implementations of the RequestWriter for HttpServletRequest will reuse
         // the existing AwsServletContext object since it has no dependencies other than the Lambda context
-        filterChainManager = new AwsFilterChainManager((AwsServletContext)context);
+        filterChainManager = new AwsFilterChainManager((AwsServletContext)servletContext);
     }
 
     protected FilterChain getFilterChain(ContainerRequestType req, Servlet servlet) {

aws-serverless-java-container-spark/src/main/java/com/amazonaws/serverless/proxy/spark/SparkLambdaContainerHandler.java

@@ -160,12 +160,16 @@ protected void handleRequest(AwsProxyHttpServletRequest httpServletRequest, AwsH
 
         if (embeddedServer == null) {
             log.debug("First request, getting new server instance");
-            embeddedServer = LambdaEmbeddedServerFactory.getServerInstance();
+
+            // trying to call init in case the embedded server had not been initialized.
+            Spark.init();
 
             // adding this call to make sure that the framework is fully initialized. This should address a race
             // condition and solve GitHub issue #71.
             Spark.awaitInitialization();
 
+            embeddedServer = LambdaEmbeddedServerFactory.getServerInstance();
+
             // call the onStartup event if set to give developers a chance to set filters in the context
             if (startupHandler != null) {
                 startupHandler.onStartup(getServletContext());