Swap Plotly prediction plot for matplotlib to rule out render freeze

Users reported the page freezing after inference on the default example.
Server-side inference completes fine in the logs, so the freeze is in the
browser rendering. Gradio 6.x has several reported issues with gr.Plot +
Plotly figures (heavy CUSTOM_CSS, visibility toggles) that match the
symptom. Switching to the existing matplotlib figure removes Plotly from
the suspect list; if the freeze persists we look elsewhere.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

app.py

@@ -940,14 +940,12 @@ def predict(sequence: str, stride: int = DEFAULT_STRIDE, threshold: float = DEFA
     # User-facing coordinates are 1-based. Core inference stays 0-based.
     display_positions = [pos + 1 for pos in result.positions]
 
-    # Create interactive Plotly plot
-    fig = create_interactive_prediction_plot(display_positions, result.probabilities, threshold, regions)
-
-    # Create static matplotlib plot for PNG/PDF export
+    # Use matplotlib figure for display AND export.
+    # Plotly + Gradio 6.x + heavy CUSTOM_CSS was freezing the browser after inference;
+    # matplotlib is the boring-and-reliable fallback to rule that out.
     output_dir = make_output_dir("crispr_prediction")
-    static_fig = create_prediction_plot(display_positions, result.probabilities, threshold, regions)
-    png_path, pdf_path = save_figure_to_file(static_fig, "crispr_prediction", output_dir)
-    plt.close(static_fig)
+    fig = create_prediction_plot(display_positions, result.probabilities, threshold, regions)
+    png_path, pdf_path = save_figure_to_file(fig, "crispr_prediction", output_dir)
 
     # Create CSV with prediction data
     csv_path = os.path.join(output_dir, "crispr_predictions.csv")