feat(loop): wire speak tool into the loop (M57)

- speak_tool.go: the speak tool itself — rune-derived timeout budget, structured
  failure codes, persona/emotion stamped on the synthesis ctx, audio persisted
  through the store seam, structured result for the model.
- voice_deps.go: the wiring seam (VoiceSynth / VoiceStore + VoiceStoreFunc
  adapter), voiceEnabled, withSpeakSpec (appended after every schema shrink),
  and captureVoice (success beats failure, last one wins).
- options.go / loop.go: VoiceSynth, VoiceStore, SpeakBudget injection points.
- request.go: advertise speak on both toolSchemas return paths.
- tools.go / tools_invoke.go: carry the batch voice outcome; split invokeOne /
  emitToolExecution / spanStatusForToolPhase / invokeParallel into their own
  file (tools.go was at the 500-line ceiling) and let speak run without a
  dispatcher.
- run.go: keep the voice outcome across tool batches (a later batch without
  speak must not erase it).

Known, deliberate: speak bypasses the capacity allow-set and skill-round
narrowing (it belongs to no built-in toolset), so it is also advertised in
quiet/awake; execution still passes the tool security gate, where speak is not
registered and non-owner actors are denied fail-closed.

internal/runtime/loop/request.go

@@ -87,7 +87,7 @@ func (l *Loop) toolSchemas(ctx context.Context, messages []model.Message) []mode
 	}
 	schemas := d.ToolSchemas()
 	if l.capacity == nil || !l.capacity.Enabled() {
-		return l.narrowToolSchemas(ctx, l.restrictSchemas(ctx, d, schemas))
+		return l.withSpeakSpec(l.narrowToolSchemas(ctx, l.restrictSchemas(ctx, d, schemas)))
 	}
 
 	names := make([]string, 0, len(schemas))
@@ -124,7 +124,7 @@ func (l *Loop) toolSchemas(ctx context.Context, messages []model.Message) []mode
 			out = append(out, s)
 		}
 	}
-	return l.narrowToolSchemas(ctx, l.restrictSchemas(ctx, d, out))
+	return l.withSpeakSpec(l.narrowToolSchemas(ctx, l.restrictSchemas(ctx, d, out)))
 }
 
 // capacityTexts 为 capacity 判定器提取近期对话文本：