Add tag-targeted test runs to CI runner

ci-runner.sh now accepts optional 4th parameter for pytest markers:
  ci-runner.sh mfg abc123 claude genlab  → runs only genlab tests
  ci-runner.sh mfg abc123 claude heavy   → runs full heavy suite
  ci-runner.sh mfg abc123 claude         → runs full light suite (default)

Changes:
- TAG parameter parsed from $4
- pytest -m flag applied when tag provided
- RUN_MODE shown in Telegram notification (light/heavy/tag:name)
- Baseline check only enforced on full light suite runs
- Comments noting tags need periodic updates as features change

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

ci-runner.sh

@@ -3,6 +3,17 @@
 # Runs tests, feature manifests, and package checks against TSHARPS worktrees.
 # READ-ONLY — never writes, commits, or pushes to TSHARPS.
 # Same checks for ALL branches — no branch-specific logic.
+#
+# Usage:
+#   ci-runner.sh <branch> <commit> <actor> [tag]
+#
+# Optional [tag] parameter runs only tests with that pytest marker:
+#   ci-runner.sh mfg abc123 claude genlab    → runs only genlab-tagged tests
+#   ci-runner.sh mfg abc123 claude           → runs full light suite (default)
+#   ci-runner.sh mfg abc123 claude heavy     → runs full heavy suite
+#
+# NOTE: Feature tags (genlab, scheduling, auth, etc.) need updating from time
+# to time as features change. See pytest.ini for the current list of markers.
 
 set -o pipefail
 
@@ -11,6 +22,7 @@ CONFIG="$SCRIPT_DIR/ci-config.json"
 BRANCH="$1"
 COMMIT="$2"
 ACTOR="${3:-unknown}"
+TAG="${4:-}"
 START_TIME=$(date +%s)
 
 # Load config
@@ -83,7 +95,21 @@ if $PYTHON -c "import pytest_timeout" 2>/dev/null; then
   TIMEOUT_FLAG="--timeout=$TIMEOUT"
 fi
 
-TEST_OUTPUT=$($PYTHON -m pytest "$WORKTREE/backend/tests/" --tb=line -q $TIMEOUT_FLAG 2>&1)
+# Build pytest command — use tag filter if provided
+TAG_FLAG=""
+RUN_MODE="light"
+if [ -n "$TAG" ]; then
+  if [ "$TAG" = "heavy" ]; then
+    TAG_FLAG=""
+    RUN_MODE="heavy"
+    export RUN_HEAVY_TESTS=1
+  else
+    TAG_FLAG="-m $TAG"
+    RUN_MODE="tag:$TAG"
+  fi
+fi
+
+TEST_OUTPUT=$($PYTHON -m pytest "$WORKTREE/backend/tests/" --tb=line -q $TIMEOUT_FLAG $TAG_FLAG 2>&1)
 TEST_EXIT=$?
 
 PASS_COUNT=$(echo "$TEST_OUTPUT" | grep -oP '\d+ passed' | grep -oP '\d+' || echo 0)
@@ -115,8 +141,9 @@ if [ "$ERROR_COUNT" != "0" ] && [ "$ERROR_COUNT" != "" ]; then
 fi
 
 # ─── Test Count Baseline Check ───
+# Baseline check only applies to full suite runs (light), not tag-targeted runs
 BASELINE=$(python3 -c "import json; print(json.load(open('$CONFIG')).get('test_count_baseline', 0))")
-if [ "$BASELINE" -gt 0 ] && [ "$PASS_COUNT" -lt "$BASELINE" ]; then
+if [ "$RUN_MODE" = "light" ] && [ "$BASELINE" -gt 0 ] && [ "$PASS_COUNT" -lt "$BASELINE" ]; then
   OVERALL="fail"
   echo "TEST COUNT REGRESSION: expected >= $BASELINE passed, got $PASS_COUNT"
 fi
@@ -151,7 +178,13 @@ SUMMARY="${PASS_COUNT} passed, ${FAIL_COUNT} failed, ${SKIP_COUNT} skipped"
 TAGS_USED=$(grep -rhoP 'pytest\.mark\.\K\w+' "$WORKTREE/backend/tests/test_"*.py 2>/dev/null | \
   sort -u | grep -vxE 'skipif|skip|parametrize|fixture|unit|integration|pipeline|e2e|slow|heavy' | \
   tr '\n' ', ' | sed 's/,$//')
-SUITE_INFO="Tags: ${TAGS_USED} | ${DESELECTED_COUNT} not in scope"
+if [ "$RUN_MODE" = "light" ]; then
+  SUITE_INFO="Mode: light (full) | Tags: ${TAGS_USED} | ${DESELECTED_COUNT} not in scope"
+elif [ "$RUN_MODE" = "heavy" ]; then
+  SUITE_INFO="Mode: heavy (all tests) | Tags: ${TAGS_USED}"
+else
+  SUITE_INFO="Mode: ${RUN_MODE} | ${DESELECTED_COUNT} not in scope"
+fi
 
 echo ""
 echo "========================="