[ML] File upload fixing inference timeout check (elastic#204722)

Fixes issue in serverless where the inference call to deploy elser times
out, which is expected, but returns with an unexpected `500` error code.
This causes the UI to stop the upload process.

![image](https://github.com/user-attachments/assets/57b3007f-5a67-4c99-b6bd-360c7b5b788f)


This PR increases the timeout length to 10mins and allows for `500`
errors just in case.
Testing this shows that with the extra timeout time in kibana, another
timeout happens at 1min but this returns with an expected `408`.
Therefore allowing for `500`s might not be needed, but I don't think
there is any harm.

x-pack/platform/plugins/private/data_visualizer/public/application/file_data_visualizer/components/import_view/auto_deploy.ts

@@ -26,7 +26,7 @@ export class AutoDeploy {
       // we'll know when it's ready from polling the inference endpoints
       // looking for num_allocations
       const status = e.response?.status;
-      if (status === 408 || status === 504 || status === 502) {
+      if (status === 408 || status === 504 || status === 502 || status === 500) {
         return;
       }
       this.inferError = e;

x-pack/platform/plugins/private/data_visualizer/server/routes.ts

@@ -145,10 +145,13 @@ export function routes(coreSetup: CoreSetup<StartDeps, unknown>, logger: Logger)
           const inferenceId = request.params.inferenceId;
           const input = request.body.input;
           const esClient = (await context.core).elasticsearch.client;
-          const body = await esClient.asCurrentUser.inference.inference({
-            inference_id: inferenceId,
-            input,
-          });
+          const body = await esClient.asCurrentUser.inference.inference(
+            {
+              inference_id: inferenceId,
+              input,
+            },
+            { maxRetries: 0, requestTimeout: 10 * 60 * 1000 }
+          );
 
           return response.ok({ body });
         } catch (e) {