feat: scan results panel, model switching, batch scan, and training improvements

- Replace librosa with direct ffmpeg subprocess for 10x faster audio loading
- Add ScanResultsPanel with per-model tabs, seek-on-click, delete, and export
- Persist scan results in DB per (filename, profile, model)
- Add model selector dropdown to switch between trained embedding models
- Add "Scan All" button for batch scanning playlist videos
- Support manual negative examples via negative class folder
- Configurable auto-negative margin (default 30s, 0 = disabled)
- Deduplicate nearby training markers (8s min gap)
- Parallel audio loading with ThreadPoolExecutor during training
- Progress callbacks from training for UI status updates
- Cache bypass in scan_video (skip audio loading when embeddings cached)
- Move all caches (models, embeddings, downloads) into project directory
- Add 8cut.sh launcher script with auto venv/conda detection
- Fix 11 bugs across thread safety, signal handling, and state management

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

8cut.sh

@@ -0,0 +1,29 @@
+#!/bin/bash
+# Launch 8-cut with auto-detected venv/conda environment
+SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd)"
+ENV_NAME="8cut"
+CONDA_PREFIX_BASE="/media/p5/miniforge3"
+
+# 1. Try .venv in project dir
+if [ -f "$SCRIPT_DIR/.venv/bin/activate" ]; then
+    source "$SCRIPT_DIR/.venv/bin/activate"
+    exec python "$SCRIPT_DIR/main.py" "$@"
+fi
+
+# 2. Try conda env (works without shell init)
+CONDA_PYTHON="$CONDA_PREFIX_BASE/envs/$ENV_NAME/bin/python"
+if [ -x "$CONDA_PYTHON" ]; then
+    exec "$CONDA_PYTHON" "$SCRIPT_DIR/main.py" "$@"
+fi
+
+# 3. Try conda via shell hook (interactive shells)
+if command -v conda &>/dev/null; then
+    eval "$(conda shell.bash hook 2>/dev/null)"
+    if conda env list 2>/dev/null | grep -qw "$ENV_NAME"; then
+        conda activate "$ENV_NAME"
+        exec python "$SCRIPT_DIR/main.py" "$@"
+    fi
+fi
+
+# 4. Fallback to system Python
+exec python3 "$SCRIPT_DIR/main.py" "$@"