springai/multimodel: capture previous response before signaling

Same bug pattern that RagApplication had: the polling loop compared
against an empty-string baseline, so the very first poll could
return the stale prior response and print it as if it were the new
one. Capture previousResponse before sending the chat signal and
wait until getLastResponse() differs from that pre-signal baseline.

Also drop the redundant initial Thread.sleep(100) — the loop's own
sleep handles backoff, and reading immediately is fine when we're
comparing against the pre-signal baseline.

Addresses Copilot review on PR #775.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

Author: donald-pinckney

springai/multimodel/src/main/java/io/temporal/samples/springai/multimodel/MultiModelApplication.java

@@ -117,19 +117,17 @@ public void run(String... args) throws Exception {
         continue;
       }
 
+      // Capture current response BEFORE sending so we can detect the change against
+      // the pre-signal baseline (not against an empty string — the workflow may already
+      // hold the previous prompt's reply).
+      String previousResponse = workflowStub.getLastResponse();
       System.out.println("[Sending to " + modelName + " model...]");
-
-      // Send the message via signal
       workflowStub.chat(modelName, message);
 
-      // Wait a moment for processing, then query for response
-      Thread.sleep(100);
-
-      // Poll for response (in production, you'd use a more sophisticated approach)
-      String lastResponse = "";
+      // Poll until the response changes from the pre-signal baseline.
       for (int i = 0; i < 300; i++) { // Wait up to 30 seconds
         String response = workflowStub.getLastResponse();
-        if (!response.isEmpty() && !response.equals(lastResponse)) {
+        if (!response.equals(previousResponse)) {
           System.out.println(
               "\n[" + modelName.toUpperCase(java.util.Locale.ROOT) + "]: " + response + "\n");
           break;