fix(docker): make bind-mounted data folders writable on startup

The image switched to the node user (uid 1000) at build time, so a host
folder mounted at /data kept its host ownership and SQLite failed with
SQLITE_CANTOPEN in a restart loop. Named and anonymous volumes worked only
because Docker copies the image's /data ownership into them.

The container now starts as root, hands DATA_DIR and the database folder
to node (only entries not already owned by it), and execs the app via
setpriv as node. Containers started with an explicit `user:` skip this.

Adds a CI step that runs the image with a host-owned bind mount.

Fixes #93

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_016MHECwBhBpM248EFSBi4L5

crumbs-entrypoint.sh

@@ -0,0 +1,20 @@
+#!/bin/sh
+# Starts as root so a bind-mounted data folder owned by the host user can be
+# handed to the unprivileged `node` user, then drops privileges before running
+# the app. Without this, a host folder mounted at /data is not writable by
+# uid 1000 and SQLite fails with SQLITE_CANTOPEN.
+set -e
+
+if [ "$(id -u)" = "0" ]; then
+	for dir in "${DATA_DIR:-/data}" "$(dirname "${DATABASE_URL:-/data/crumbs.db}")"; do
+		mkdir -p "$dir"
+		# Only touch entries not already owned by node, so restarts on a large
+		# attachment folder stay fast.
+		find "$dir" \! -user node -exec chown node:node {} + ||
+			echo "warning: could not change ownership of $dir; it must be writable by uid 1000" >&2
+	done
+	exec setpriv --reuid=node --regid=node --init-groups "$@"
+fi
+
+# Started with an explicit non-root user (e.g. `user:` in compose): run as-is.
+exec "$@"