Add ephemeral turn context support

Introduce app-server ephemeralContext, render it as additional_context_for_this_turn fragments, and strip it correctly across contextual-user diffing and compaction flows.

Regenerate app-server schema fixtures and add coverage for model-visible layout, compaction, oversize validation, and resumed/forked context reconstruction.

Co-authored-by: Codex <noreply@openai.com>

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

@@ -90,6 +90,23 @@
       ],
       "type": "object"
     },
+    "EphemeralContext": {
+      "properties": {
+        "text": {
+          "description": "Free-form text payload for additional context sent with one turn.",
+          "type": "string"
+        },
+        "title": {
+          "description": "Human-readable title for additional context sent with one turn.",
+          "type": "string"
+        }
+      },
+      "required": [
+        "text",
+        "title"
+      ],
+      "type": "object"
+    },
     "ModeKind": {
       "description": "Initial collaboration mode to use when the TUI starts.",
       "enum": [
@@ -520,6 +537,16 @@
       ],
       "description": "Override the reasoning effort for this turn and subsequent turns."
     },
+    "ephemeralContext": {
+      "description": "Additional model-visible context for this turn, such as editor or IDE state. This context is not re-injected automatically after compaction, so clients should send a fresh snapshot on each turn instead of expecting it to carry forward automatically.",
+      "items": {
+        "$ref": "#/definitions/EphemeralContext"
+      },
+      "type": [
+        "array",
+        "null"
+      ]
+    },
     "input": {
       "items": {
         "$ref": "#/definitions/UserInput"