Stage 5: Alembic migrations, durable call rows, one transcript truth

Alembic replaces create_all as the schema authority: async env.py
against Base.metadata (CLI and in-app entry paths share it via
config.attributes["connection"]), an autogenerated baseline of the
create_all-era schema, and init_db now runs upgrade head — stamping
the baseline first on a pre-Alembic database so existing deployments
adopt cleanly. create_all remains for tests only.

Calls are durable from the start: CallManager gains an
on_call_created hook (wired to persist_call_on_create) that inserts
an in_progress CallRecord the moment a call is created;
persist_call_on_end finalizes that same row. A SIGKILL mid-call now
leaves an in_progress row instead of erasing the call from history
(verified live against the dev database).

One transcript representation: ActiveCall.transcript_chunks holds
TranscriptEntry (t_offset_ms, speaker, text) — add_transcript stamps
real offsets from connect time, receptionist passes speaker instead
of encoding it into "caller: ..." strings, persisted chunks carry
real seek offsets, and the dead CallRecord.transcript Text column is
dropped by migration. Device.is_online migrates String → Boolean
(with a USING cast for existing rows).

Model de-triplication: CallResponse/CallStatusResponse build via
from_call classmethods (one ActiveCall→response mapping);
DeviceStatus deleted — can_receive_call is a computed field on
Device and the list endpoint returns the domain model; all row↔dict
and row↔domain mapping now lives in call_persistence.py
(record_summary/record_detail/chunk_to_dict + device row functions).

New tests/test_data_layer.py: upgrade-head-matches-models,
pre-Alembic adoption, durable in_progress rows, end-without-create
fallback, transcript offsets, consolidated response models.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

core/call_manager.py

@@ -5,15 +5,19 @@ Central nervous system of the gateway. Tracks all active calls,
 publishes events, and coordinates between SIP engine and services.
 """
 
-import asyncio
 import logging
 import uuid
-from collections.abc import AsyncIterator
 from datetime import datetime
 from typing import Optional
 
-from core.event_bus import EventBus, EventSubscription
-from models.call import ActiveCall, AudioClassification, CallMode, CallStatus, ClassificationResult
+from core.event_bus import EventBus
+from models.call import (
+    ActiveCall,
+    CallMode,
+    CallStatus,
+    ClassificationResult,
+    TranscriptEntry,
+)
 from models.events import EventType, GatewayEvent
 
 logger = logging.getLogger(__name__)
@@ -26,10 +30,11 @@ class CallManager:
     The single source of truth for what's happening on the gateway.
     """
 
-    def __init__(self, event_bus: EventBus, on_call_ended=None):
+    def __init__(self, event_bus: EventBus, on_call_created=None, on_call_ended=None):
         self.event_bus = event_bus
         self._active_calls: dict[str, ActiveCall] = {}
         self._call_legs: dict[str, str] = {}  # SIP leg ID -> call ID mapping
+        self._on_call_created = on_call_created  # async callback(call)
         self._on_call_ended = on_call_ended  # async callback(call, final_status)
 
     # ================================================================
@@ -67,6 +72,14 @@ class CallManager:
             message=f"📞 Calling {remote_number} ({mode.value})",
         ))
 
+        # Durable in_progress row — a crash mid-call must not erase the
+        # call from history. The hook does its own retrying/logging.
+        if self._on_call_created is not None:
+            try:
+                await self._on_call_created(call)
+            except Exception as e:
+                logger.warning(f"on_call_created hook failed for {call_id}: {e}")
+
         return call
 
     async def update_status(self, call_id: str, status: CallStatus) -> None:
@@ -135,18 +148,26 @@ class CallManager:
             message=f"🎵 Audio: {result.audio_type.value} ({result.confidence:.0%})",
         ))
 
-    async def add_transcript(self, call_id: str, text: str) -> None:
-        """Add a transcript chunk to a call."""
+    async def add_transcript(
+        self, call_id: str, text: str, speaker: str = "unknown"
+    ) -> None:
+        """Add a transcript entry to a call, stamped with its offset."""
         call = self._active_calls.get(call_id)
         if not call:
             return
 
-        call.transcript_chunks.append(text)
+        anchor = call.connected_at or call.started_at
+        entry = TranscriptEntry(
+            t_offset_ms=int((datetime.now() - anchor).total_seconds() * 1000),
+            speaker=speaker,
+            text=text,
+        )
+        call.transcript_chunks.append(entry)
 
         await self.event_bus.publish(GatewayEvent(
             type=EventType.TRANSCRIPT_CHUNK,
             call_id=call_id,
-            data={"text": text},
+            data={"text": text, "speaker": speaker, "t_offset_ms": entry.t_offset_ms},
             message=f"📝 '{text[:80]}...' " if len(text) > 80 else f"📝 '{text}'",
         ))