convert wd_test to use sh_test

build/fixtures/BUILD.bazel

@@ -1 +1,4 @@
-exports_files(["kj_test.sh"])
+exports_files([
+    "kj_test.sh",
+    "wd_test.sh",
+])

build/fixtures/wd_test.sh

@@ -0,0 +1,46 @@
+#!/bin/bash
+# Test runner script for wd_test. Handles three modes:
+# - Normal: runs workerd test directly
+# - Sidecar: runs workerd test via supervisor which manages a parallel sidecar process
+# - Snapshot: runs workerd twice (save snapshot, then load snapshot) for Python tests
+#
+# Environment variables (set by wd_test.bzl):
+#   SIDECAR_COMMAND     - If set, enables sidecar mode
+#   SIDECAR_SUPERVISOR  - Path to supervisor executable (required for sidecar mode)
+#   PORTS_TO_ASSIGN     - Comma-separated port binding names for sidecar
+#   RANDOMIZE_IP        - "true"/"false" for sidecar IP randomization
+#   PYTHON_SNAPSHOT_TEST - If set, enables snapshot mode
+#   PYTHON_SAVE_SNAPSHOT_ARGS - Additional args for snapshot save phase
+
+set -euo pipefail
+
+# Set up coverage for workerd subprocess.
+# LLVM_PROFILE_FILE tells the coverage-instrumented workerd binary where to write profraw data.
+# KJ_CLEAN_SHUTDOWN ensures proper shutdown so coverage data is flushed.
+if [ -n "${COVERAGE_DIR:-}" ]; then
+    export LLVM_PROFILE_FILE="$COVERAGE_DIR/%p.profraw"
+    export KJ_CLEAN_SHUTDOWN=1
+fi
+
+run_test() {
+    "$@" -dTEST_TMPDIR="${TEST_TMPDIR:-/tmp}"
+}
+
+# Sidecar mode: run via supervisor which starts sidecar process in parallel.
+# The supervisor handles port assignment and ensures sidecar is ready before test starts.
+if [ -n "${SIDECAR_COMMAND:-}" ]; then
+    "$SIDECAR_SUPERVISOR" run_test "$@"
+
+# Snapshot mode: run workerd twice for Python memory snapshot tests.
+# First run creates the snapshot, second run uses it.
+elif [ -n "${PYTHON_SNAPSHOT_TEST:-}" ]; then
+    echo "Creating Python Snapshot"
+    run_test "$@" --python-save-snapshot ${PYTHON_SAVE_SNAPSHOT_ARGS:-}
+    echo ""
+    echo "Using Python Snapshot"
+    run_test "$@" --python-load-snapshot snapshot.bin
+
+# Normal mode: just run the test.
+else
+    run_test "$@"
+fi