Add back keep_alive to async_search.submit (#3601) (#3807)

(cherry picked from commit c3c2156)

# Conflicts:
#	output/schema/schema-serverless.json
#	output/schema/schema.json
#	output/schema/validation-errors.json

Author: pquentin

output/openapi/elasticsearch-openapi.json

@@ -2,6 +2,7 @@
   "endpointErrors": {
     "async_search.submit": {
       "request": [
+        "Request: query parameter 'keep_alive' does not exist in the json spec",
         "Request: query parameter 'min_compatible_shard_node' does not exist in the json spec",
         "interface definition _types:QueryVectorBuilder - Property text_embedding is a single-variant and must be required"
       ],

output/openapi/elasticsearch-serverless-openapi.json

@@ -68,7 +68,7 @@ import { Duration } from '@_types/Time'
  * @doc_tag search
  */
 // NOTE: this is a SearchRequest with:
-//  * 2 added parameters: wait_for_completion_timeout, keep_on_completion
+//  * 2 added parameters: wait_for_completion_timeout, keep_on_completion and keep_alive
 //  * 2 removed parameters: scroll, pre_filter_shard_size
 export interface Request extends RequestBase {
   urls: [
@@ -91,6 +91,12 @@ export interface Request extends RequestBase {
      * @server_default 1s
      */
     wait_for_completion_timeout?: Duration
+    /**
+     * Specifies how long the async search needs to be available.
+     * Ongoing async searches and any saved search results are deleted after this period.
+     * @server_default 5d
+     */
+    keep_alive?: Duration
     /**
      * If `true`, results are stored for later retrieval when the search completes within the `wait_for_completion_timeout`.
      * @server_default false