Raise Modal generation timeout to 2 hours

A 120-min+ record attempt hit the 3600s ceiling and lost the entire
render. The generator streams progress throughout, so a long timeout is
insurance against losing an hour of GPU work, not a runaway risk. Clone
references also prefill slower than the preset-voice pace the old
ceiling was sized against.

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

backend_modal/modal_runner.py

@@ -65,7 +65,10 @@ cache_volume = modal.Volume.from_name("vibevoice-cache", create_if_missing=True)
 @app.cls(
     gpu="A100-40GB",
     scaledown_window=300,
-    timeout=3600,  # long-form renders (90-min scripts ~25 min wall) need headroom
+    timeout=7200,  # was 3600: a 120-min+ record attempt hit the 1h ceiling at the
+                   # finish line (2026-08-19), losing the whole render. Long-form
+                   # with cloned voices runs slower than the preset-voice record
+                   # pace (bigger reference prefill per chunk), so give 2h.
     volumes={"/cache": cache_volume}
 )
 class VibeVoiceModel: