Persist timestamped realtime frames before acknowledging completion (#61473)

## What changed

- Record voice transcripts as `frame` items with a `voice_transcript` channel for eligible threads, while retaining legacy transcript support and the existing app-server live and mixed timeline representation.
- Assign first-observed timestamps that never move backward, retain them through completion, and restore the clock floor from history and compaction checkpoints.
- Flush realtime records before completion notifications, including file and initial directory durability barriers for local rollouts.
- Generate a distinct default session ID for each new call, use the call ID when attaching to an existing call, and derive opening record IDs from session IDs.
- Add protocol types for turn-input acceptance and session-end input receipts, excluding those receipts from live notifications and mixed timelines. Update generated schemas and SDK types.

## Testing

Add coverage for frame serialization and required voice payloads, clock rollback and checkpoint restoration, durability barrier failures and retries, per-call session IDs, and persisted transcript availability immediately after completion.

GitOrigin-RevId: 3f6632b517d563861aee085b723475b8945c391c

codex-rs/app-server-protocol/schema/json/v2/ThreadRealtimeItemStartedNotification.json

@@ -60,6 +60,25 @@
         }
       ]
     },
+    "ThreadRealtimeFrameChannels": {
+      "description": "A frame holds one or more payloads, at most one per channel. They share one ID, observation time, and seal. V1 requires its only channel, `voice_transcript`.",
+      "properties": {
+        "voice_transcript": {
+          "$ref": "#/definitions/ThreadRealtimeVoiceTranscript"
+        }
+      },
+      "required": [
+        "voice_transcript"
+      ],
+      "type": "object"
+    },
+    "ThreadRealtimeHandoffMode": {
+      "enum": [
+        "start",
+        "steer"
+      ],
+      "type": "string"
+    },
     "ThreadRealtimeItem": {
       "description": "EXPERIMENTAL - a thread-scoped realtime item in the canonical timeline.",
       "oneOf": [
@@ -79,6 +98,26 @@
           "title": "RealtimeSessionStartedThreadRealtimeItem",
           "type": "object"
         },
+        {
+          "properties": {
+            "channels": {
+              "$ref": "#/definitions/ThreadRealtimeFrameChannels"
+            },
+            "type": {
+              "enum": [
+                "frame"
+              ],
+              "title": "FrameThreadRealtimeItemType",
+              "type": "string"
+            }
+          },
+          "required": [
+            "channels",
+            "type"
+          ],
+          "title": "FrameThreadRealtimeItem",
+          "type": "object"
+        },
         {
           "properties": {
             "role": {
@@ -150,12 +189,94 @@
           ],
           "title": "RealtimeSessionClosedThreadRealtimeItem",
           "type": "object"
+        },
+        {
+          "properties": {
+            "handoff_id": {
+              "type": "string"
+            },
+            "input": {
+              "type": "string"
+            },
+            "input_item_id": {
+              "type": "string"
+            },
+            "mode": {
+              "$ref": "#/definitions/ThreadRealtimeHandoffMode"
+            },
+            "source_event_ids": {
+              "items": {
+                "type": "string"
+              },
+              "type": "array"
+            },
+            "turn_id": {
+              "type": "string"
+            },
+            "type": {
+              "enum": [
+                "turnInputAccepted"
+              ],
+              "title": "TurnInputAcceptedThreadRealtimeItemType",
+              "type": "string"
+            }
+          },
+          "required": [
+            "handoff_id",
+            "input",
+            "input_item_id",
+            "mode",
+            "source_event_ids",
+            "turn_id",
+            "type"
+          ],
+          "title": "TurnInputAcceptedThreadRealtimeItem",
+          "type": "object"
+        },
+        {
+          "properties": {
+            "input": {
+              "type": "string"
+            },
+            "input_item_id": {
+              "type": "string"
+            },
+            "source_event_ids": {
+              "items": {
+                "type": "string"
+              },
+              "type": "array"
+            },
+            "type": {
+              "enum": [
+                "sessionEndInputSaved"
+              ],
+              "title": "SessionEndInputSavedThreadRealtimeItemType",
+              "type": "string"
+            }
+          },
+          "required": [
+            "input",
+            "input_item_id",
+            "source_event_ids",
+            "type"
+          ],
+          "title": "SessionEndInputSavedThreadRealtimeItem",
+          "type": "object"
         }
       ],
       "properties": {
         "id": {
           "type": "string"
         },
+        "observedAtMs": {
+          "description": "Canonical observation time in Unix milliseconds for eligible voice calls.",
+          "format": "int64",
+          "type": [
+            "integer",
+            "null"
+          ]
+        },
         "realtimeSessionId": {
           "type": "string"
         }
@@ -179,6 +300,21 @@
         "assistant"
       ],
       "type": "string"
+    },
+    "ThreadRealtimeVoiceTranscript": {
+      "properties": {
+        "role": {
+          "$ref": "#/definitions/ThreadRealtimeTranscriptRole"
+        },
+        "text": {
+          "type": "string"
+        }
+      },
+      "required": [
+        "role",
+        "text"
+      ],
+      "type": "object"
     }
   },
   "description": "EXPERIMENTAL - a realtime timeline item started before its content streams.",