Add grace period and retry logic for Fabric Card firmware upload

After an IOM reboot, the Redfish API responds before inter-IOM services
are fully initialized. This causes the Fabric Card upload to fail with
"Failed to send update package to other IOM; Couldn't connect to server".

Two fixes:
1. After _wait_for_iom_online confirms the IOM is back, wait an additional
   60s for services to fully initialize before proceeding.
2. Retry the Fabric Card firmware upload up to 3 times with 60s between
   attempts, covering cases where services are still starting up.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

modules/workflow_firmware.py

@@ -166,6 +166,8 @@ def _update_iom_fw(password: str, ip: str, iom: str, fw_path: str) -> bool:
     time.sleep(30)  # allow time for the IOM to begin shutting down before polling
     if _wait_for_iom_online(password, ip):
         ok(f"{iom} is back online.")
+        info("Allowing IOM services to fully initialize before next step...")
+        time.sleep(60)
     else:
         warn(f"{iom} did not respond within 5 minutes — proceeding anyway.")
     return True
@@ -179,10 +181,25 @@ def _update_fabric_fw(password: str, ip: str, iom: str, fw_path: str) -> bool:
     After the update: restart fabric card, then restart IOM.
     """
     sz = os.path.getsize(fw_path)
-    info(f"Uploading Fabric Card firmware ({sz // 1024} KB) to {iom} at {ip}...")
+
-    ok_flag, data = _redfish_upload_firmware(password, ip, fw_path)
+    # Retry the upload — after an IOM reboot the inter-IOM services can take
+    # time to initialize, causing the first upload attempt to fail with
+    # "Failed to send update package to other IOM".
+    MAX_UPLOAD_ATTEMPTS = 3
+    ok_flag, data = False, ""
+    for attempt in range(1, MAX_UPLOAD_ATTEMPTS + 1):
+        info(f"Uploading Fabric Card firmware ({sz // 1024} KB) to {iom} at {ip}"
+             + (f" (attempt {attempt}/{MAX_UPLOAD_ATTEMPTS})" if attempt > 1 else "") + "...")
+        ok_flag, data = _redfish_upload_firmware(password, ip, fw_path)
+        if ok_flag:
+            break
+        if attempt < MAX_UPLOAD_ATTEMPTS:
+            warn(f"Upload failed: {data}")
+            info("Waiting 60s for IOM services to finish initializing...")
+            time.sleep(60)
+
     if not ok_flag:
-        error(f"Upload failed: {data}")
+        error(f"Upload failed after {MAX_UPLOAD_ATTEMPTS} attempts: {data}")
         return False
     ok("Firmware file uploaded.")