docs(graph): Add DSL documentation and examples

Add comprehensive documentation for the graph-based pipeline DSL:

- docs/graph-dsl.md: Complete DSL reference with TOML, Python, and CLI syntax
- docs/GRAPH_SYSTEM_SUMMARY.md: Implementation overview and architecture
- examples/graph_dsl_demo.py: Demonstrates imperative Python API usage
- examples/test_graph_integration.py: Integration test verifying pipeline execution

The documentation follows a wiki-like approach with navigable structure:
- Overview section explaining the concept
- Syntax examples for each format (TOML, Python, CLI)
- Node type reference table
- Advanced features section
- Comparison with old XYZStage approach

This provides users with multiple entry points to understand and use the
new graph-based pipeline system.

examples/test_graph_integration.py

@@ -0,0 +1,110 @@
+#!/usr/bin/env python3
+"""
+Test script to verify graph-based pipeline integration.
+
+This script tests that the graph DSL can be used to create working pipelines
+that produce output similar to preset-based pipelines.
+"""
+
+from engine.effects.plugins import discover_plugins
+from engine.pipeline.graph_toml import load_pipeline_from_toml
+from engine.pipeline.params import PipelineParams
+
+
+def test_graph_pipeline_execution():
+    """Test that a graph-based pipeline can execute and produce output."""
+    print("=== Testing Graph Pipeline Execution ===")
+
+    # Discover plugins
+    discover_plugins()
+
+    # Load pipeline from TOML
+    pipeline = load_pipeline_from_toml(
+        "examples/pipeline_graph.toml", viewport_width=80, viewport_height=24
+    )
+
+    print(f"Pipeline loaded with {len(pipeline._stages)} stages")
+    print(f"Stages: {list(pipeline._stages.keys())}")
+
+    # Initialize pipeline
+    if not pipeline.initialize():
+        print("Failed to initialize pipeline")
+        return False
+
+    print("Pipeline initialized successfully")
+
+    # Set up context
+    ctx = pipeline.context
+    params = PipelineParams(viewport_width=80, viewport_height=24)
+    ctx.params = params
+
+    # Execute pipeline with empty items (source will provide content)
+    result = pipeline.execute([])
+
+    if result.success:
+        print(f"Pipeline executed successfully")
+        print(f"Output type: {type(result.data)}")
+        if isinstance(result.data, list):
+            print(f"Output lines: {len(result.data)}")
+            if len(result.data) > 0:
+                print(f"First line: {result.data[0][:50]}...")
+        return True
+    else:
+        print(f"Pipeline execution failed: {result.error}")
+        return False
+
+
+def test_graph_vs_preset():
+    """Compare graph-based and preset-based pipelines."""
+    print("\n=== Comparing Graph vs Preset ===")
+
+    from engine.pipeline import get_preset
+
+    # Load graph-based pipeline
+    graph_pipeline = load_pipeline_from_toml(
+        "examples/pipeline_graph.toml", viewport_width=80, viewport_height=24
+    )
+
+    # Load preset-based pipeline (using test-basic as a base)
+    preset = get_preset("test-basic")
+    if not preset:
+        print("test-basic preset not found")
+        return False
+
+    # Create pipeline from preset config
+    from engine.pipeline import Pipeline
+
+    preset_pipeline = Pipeline(config=preset.to_config())
+
+    print(f"Graph pipeline stages: {len(graph_pipeline._stages)}")
+    print(f"Preset pipeline stages: {len(preset_pipeline._stages)}")
+
+    # Compare stage types
+    graph_stage_types = {
+        name: stage.__class__.__name__ for name, stage in graph_pipeline._stages.items()
+    }
+    preset_stage_types = {
+        name: stage.__class__.__name__
+        for name, stage in preset_pipeline._stages.items()
+    }
+
+    print("\nGraph pipeline stages:")
+    for name, stage_type in graph_stage_types.items():
+        print(f"  - {name}: {stage_type}")
+
+    print("\nPreset pipeline stages:")
+    for name, stage_type in preset_stage_types.items():
+        print(f"  - {name}: {stage_type}")
+
+    return True
+
+
+if __name__ == "__main__":
+    success1 = test_graph_pipeline_execution()
+    success2 = test_graph_vs_preset()
+
+    if success1 and success2:
+        print("\n✓ All tests passed!")
+    else:
+        print("\n✗ Some tests failed")
+        exit(1)