feat(04-01): data model changes + HTML layout restructure

- Add CueModel.duration field (float = 4.0) for block length in seconds
- Add AnalysisBlock.calibration_offset field (float = 0.03) for librosa bias
- Replace main-area animation grid with timeline canvas layout (toolbar, canvas, inspector)
- Move ANIMATIONS panel into sidebar below DEVICES panel
- Add timeline-toolbar with zoom slider, beat offset input, snap toggle

lightsync/models/show.py

@@ -16,6 +16,7 @@ class AudioRef(BaseModel):
 class CueModel(BaseModel):
     id: UUID = Field(default_factory=uuid4)
     timestamp: float
+    duration: float = 4.0  # block length in seconds; defaults for old files
     mode: Literal["animation", "frame_sequence"] = "animation"
     animation: str | None = None
     params: dict[str, Any] = Field(default_factory=dict)
@@ -31,6 +32,7 @@ class AnalysisBlock(BaseModel):
     tempo_bpm: float | None = None
     beat_times: list[float] = Field(default_factory=list)
     onset_times: list[float] = Field(default_factory=list)
+    calibration_offset: float = 0.03  # seconds; compensates librosa latency bias
 
 
 class ShowModel(BaseModel):