Unhandled case in init.d/srv script

When there is no mountpoint /srv, the srv.init script will fail when doing mount /srv. Even in the 'fallback' case (the die() function), this won't work.

This should probably never happen, but after I faked a drive failure (by powering off) in my RAID1 configuration, /srv was gone, preventing my MOX from properly rebooting and remounting /srv.

A simple mkdir /srv took care of it, but this case might be considered in the init script.

---

Originally submitted by @th1j5.

I am recreating this issue after gitlab data loss.