Runtime footer: app-controlled model/context/cwd/latency/cost under replies

Hermes can append a text "runtime footer" (model, context %, workdir,
latency, cost) to final replies, but only when display.runtime_footer is
enabled in the hermes config. We want the same info but controlled by the
APP, not the gateway config. So the gateway now ALWAYS sends the data as a
structured `runtime` object on final assistant messages, and the app decides
whether/what to show.

Gateway (gateway-plugin/):
- protocol.py: new runtime_footer() helper + `runtime` field on the
  message / message.stop frames. Keys (all optional, absent when the data
  is unavailable — e.g. no cost for local models): model (vendor prefix
  dropped), context_pct (0-100), cwd (home-relative), latency (seconds),
  cost (USD).
- adapter.py: a post_api_request plugin hook captures the turn's model +
  prompt tokens + start time (platform-filtered to android so other
  platforms don't pollute the buffer). _build_runtime_footer() resolves the
  model's context window (cached, best-effort, off the event loop via
  asyncio.to_thread with a timeout) and computes context_pct. The runtime
  object is attached on every final send (streaming message.stop and
  non-streaming message, plus the fallback paths).
- outbox.py: `runtime` preserved in history reconstruction so the footer
  survives a restart / first open.

App (app/shared/):
- Protocol.kt: RuntimeMeta data class + `runtime` on MessagePayload /
  MessageStopPayload / HistoryMessage.
- ChatStore.kt: `runtime` on MessageItem, wired through live + history
  reconciliation.
- SecureStore.kt (+ Android/Desktop actuals): runtimeFooterEnabled +
  runtimeFooterFields (persisted per device).
- IrisController.kt: StateFlows + toggleRuntimeFooter() /
  toggleRuntimeField(); RUNTIME_FIELD_KEYS / default set / parser.
- SettingsScreen.kt: "Runtime footer" switch; when on, an expandable chip
  menu (Model · Context % · Workdir · Latency · Cost) to pick fields.
- ChatScreen.kt: footer rendered on the SAME line as the timestamp (footer
  left, time right, Telegram-style), only for final non-streaming assistant
  answers; Inspector pane now shows the runtime fields too.

Docs: 04-wire-protocol.md + frames.schema.json document the `runtime`
object.

Verified end-to-end on device: final replies carry
`qwen3.8-27B-exl3-4.5bpw · 53% · ~ · 38s` with the time right-aligned on
the same line; 69/69 gateway tests pass, Kotlin builds + tests pass.

gateway-plugin/protocol.py

@@ -249,6 +249,54 @@ def hello_ack(
     )
 
 
+# ---------------------------------------------------------------------------
+# Runtime-metadata footer (app-controlled display)
+#
+# The gateway ALWAYS attaches a structured ``runtime`` object to final
+# assistant messages so the app can render a Telegram-style footer (model,
+# context %, cwd, latency, cost). Whether/what is shown is a per-app setting,
+# NOT a hermes config — the gateway sends the data unconditionally and the
+# app decides. Mirrors hermes' ``gateway/runtime_footer.py`` fields but as
+# structured data (the app formats + picks fields).
+#
+# Recognised keys (all optional; absent when the data is unavailable):
+#   model        — bare model id, vendor prefix dropped (``gpt-5.4``)
+#   context_pct  — last-call context occupancy, 0-100 (int)
+#   cwd          — home-relative working dir (``~``)
+#   latency      — wall-clock turn duration, seconds (float)
+#   cost         — turn cost, USD (float); absent for local/free models
+# ---------------------------------------------------------------------------
+
+RUNTIME_FIELDS: tuple[str, ...] = ("model", "context_pct", "cwd", "latency", "cost")
+
+
+def runtime_footer(
+    *,
+    model: str | None = None,
+    context_pct: int | None = None,
+    cwd: str | None = None,
+    latency: float | None = None,
+    cost: float | None = None,
+) -> dict[str, Any]:
+    """Build the structured ``runtime`` footer object.
+
+    Only fields with data are included (a partially-populated footer is
+    better than empty slots). Returns ``{}`` when nothing is available.
+    """
+    d: dict[str, Any] = {}
+    if model:
+        d["model"] = model
+    if context_pct is not None:
+        d["context_pct"] = max(0, min(100, int(context_pct)))
+    if cwd:
+        d["cwd"] = cwd
+    if latency is not None and latency >= 0:
+        d["latency"] = round(latency, 3)
+    if cost is not None and cost > 0:
+        d["cost"] = round(cost, 6)
+    return d
+
+
 # Frame builder mirrors the wire schema (docs/04); the many fields are the
 # message's full shape, so the arg count is intentional.
 def message(  # noqa: PLR0913
@@ -263,6 +311,7 @@ def message(  # noqa: PLR0913
     reply_to: str | None = None,
     model: str | None = None,
     tokens: int | None = None,
+    runtime: dict[str, Any] | None = None,
     ts: int | None = None,
 ) -> Frame:
     payload: dict[str, Any] = {
@@ -280,11 +329,11 @@ def message(  # noqa: PLR0913
         payload["model"] = model
     if tokens is not None:
         payload["tokens"] = tokens
+    if runtime:
+        payload["runtime"] = runtime
     if ts is not None:
         payload["ts"] = ts
-    return Frame(
-        type=TYPE_MESSAGE, chat_id=chat_id, thread_id=thread_id, payload=payload
-    )
+    return Frame(type=TYPE_MESSAGE, chat_id=chat_id, thread_id=thread_id, payload=payload)
 
 
 def typing(chat_id: str, on: bool = True, *, thread_id: str | None = None) -> Frame:
@@ -342,6 +391,7 @@ def message_stop(
     reasoning: str | None = None,
     model: str | None = None,
     tokens: int | None = None,
+    runtime: dict[str, Any] | None = None,
     ts: int | None = None,
 ) -> Frame:
     """Finalize a streaming bubble."""
@@ -355,6 +405,8 @@ def message_stop(
         payload["model"] = model
     if tokens is not None:
         payload["tokens"] = tokens
+    if runtime:
+        payload["runtime"] = runtime
     if ts is not None:
         payload["ts"] = ts
     return Frame(
@@ -648,9 +700,7 @@ def notification(
     payload: dict[str, Any] = {"kind": kind, "title": title, "body": body}
     if ts is not None:
         payload["ts"] = ts
-    return Frame(
-        type=TYPE_NOTIFICATION, chat_id=chat_id, thread_id=thread_id, payload=payload
-    )
+    return Frame(type=TYPE_NOTIFICATION, chat_id=chat_id, thread_id=thread_id, payload=payload)
 
 
 def fcm_register(fcm_token: str | None = None, ntfy_topic: str | None = None) -> Frame:
@@ -713,9 +763,7 @@ def media_offer(
     }
     if message_id:
         payload["message_id"] = message_id
-    return Frame(
-        type=TYPE_MEDIA_OFFER, chat_id=chat_id, thread_id=thread_id, payload=payload
-    )
+    return Frame(type=TYPE_MEDIA_OFFER, chat_id=chat_id, thread_id=thread_id, payload=payload)
 
 
 def media_pull_end(ok: bool, *, id: int | None = None) -> Frame: