feat: version-agnostic analysis scripts with shared version helpers

- scripts/bicorder_common.py (new): single source of truth for historical
  gradient renames (COLUMN_RENAMES), version detection (bicorder_version col
  → version col → data/<type>_<version>/ dir convention), and training-CSV
  auto-selection (find_training_csv)
- classify_readings.py: auto-select training run by recorded bicorder version
  (excludes the input itself to avoid circular training), canonicalize old
  column names, version-mismatch warnings
- bicorder_classifier.py / export_model_for_js.py: share renames + dimension
  loader; instructive error when clustering results are missing;
  bicorder_version recorded in exported models
- compare_analyses.py: CLI (reference + comparison CSVs), rename
  canonicalization so runs of different versions align on shared gradients,
  Descriptor dedup (was silently skewing merges); legacy no-arg audit intact
- scripts/univariate_analysis.py (new): per-protocol/per-gradient averages,
  distributions, summary stats; --img publishes the three README summary
  charts to img/
- sync_readings.sh: defer classifier training to auto-matching; gitignore
  analysis/.venv and __pycache__

analysis/scripts/sync_readings.sh

@@ -7,7 +7,11 @@
 #   scripts/sync_readings.sh data/manual_20260320
 #   scripts/sync_readings.sh data/manual_20260320 --no-analysis
 #   scripts/sync_readings.sh data/manual_20260320 --min-coverage 0.8
-#   scripts/sync_readings.sh data/manual_20260320 --training data/synthetic_1.2.6/readings.csv
+#   scripts/sync_readings.sh data/manual_20260320 --training data/synthetic_1.4.0/readings.csv
+#
+# By default the classifier training CSV is auto-selected to match the synced
+# dataset's recorded bicorder version (see scripts/bicorder_common.py); --training
+# overrides that.
 #
 # .sync_source format:
 #   REMOTE_URL=https://git.example.org/user/repo
@@ -18,7 +22,7 @@ set -euo pipefail
 DATASET_DIR="${1:?Usage: $0 <dataset_dir> [--no-analysis] [--min-coverage N]}"
 RUN_ANALYSIS=true
 MIN_COVERAGE=0.8
-TRAINING_CSV="data/synthetic_1.2.6/readings.csv"
+TRAINING_CSV=""  # empty → let classify_readings.py auto-match the dataset's bicorder version
 
 shift || true
 while [[ $# -gt 0 ]]; do
@@ -96,11 +100,18 @@ if [[ "$RUN_ANALYSIS" == true ]]; then
   echo "Generating LDA visualization..."
   "$PYTHON" scripts/lda_visualization.py "$DATASET_DIR/readings.csv"
 
-  echo ""
-  echo "Classifying readings (training: $TRAINING_CSV)..."
-  "$PYTHON" scripts/classify_readings.py \
-    "$DATASET_DIR/readings.csv" \
-    --training "$TRAINING_CSV"
+  if [[ -n "$TRAINING_CSV" ]]; then
+    echo ""
+    echo "Classifying readings (training: $TRAINING_CSV)..."
+    "$PYTHON" scripts/classify_readings.py \
+      "$DATASET_DIR/readings.csv" \
+      --training "$TRAINING_CSV"
+  else
+    echo ""
+    echo "Classifying readings (training auto-matched by bicorder version)..."
+    "$PYTHON" scripts/classify_readings.py \
+      "$DATASET_DIR/readings.csv"
+  fi
 fi
 
 echo ""