🐛 Fix entrypoint arg passthrough — commands no longer hang 🐛

Root cause: entrypoint.sh unconditionally called `__no_exit` (an
`exec bash -c` monitoring loop) even when the user passed a command
to `docker run`, replacing the shell before the command could execute.
Secondary bug in `__exec_command`: used `bash --login -c "$cmdExec"`
which only captured the first word of the command and spawned a slow
login shell on every invocation.
- rootfs/usr/local/bin/entrypoint.sh: only call `__no_exit` when no
user command was given (`$# -eq 0`); otherwise fall through to the
case dispatch so `docker run ... go version` works correctly
- rootfs/usr/local/etc/docker/functions/entrypoint.sh: rewrite
`__exec_command` to use `exec "$@"` directly instead of wrapping
in `bash --login -c "$cmdExec"`; both files kept in sync

rootfs/usr/local/bin/entrypoint.sh
rootfs/usr/local/etc/docker/functions/entrypoint.sh

rootfs/usr/local/bin/entrypoint.sh

@@ -506,9 +506,11 @@ if [ "$START_SERVICES" = "yes" ] || [ -z "$1" ]; then
     echo "$$" >"$ENTRYPOINT_PID_FILE"
     __start_init_scripts "/usr/local/etc/docker/init.d"
     CONTAINER_INIT="${CONTAINER_INIT:-no}"
-    # Services started successfully - enter monitoring mode
-    __no_exit
-    exit $?
+    # Only block in monitoring mode when no user command was given
+    if [ $# -eq 0 ]; then
+      __no_exit
+      exit $?
+    fi
   fi
   START_SERVICES="no"
 fi