whey /data-venus: adopt lezama's .automount shape, or route to the mount-unit owner

whey's /data-venus is a plain systemd .mount unit and is currently mounted ONLY because bin-whey-cc started it by hand on 2026-09-10 after a 3h16m outage (read: Connection reset by peer at 2026-09-09 23:09:02, a boot race). It will not survive a reboot. Two concrete defects in whey's unit: (1) it carries Restart=on-failure and RestartSec=5, which are [Service] directives and are SILENTLY IGNORED in a [Mount] section - those two lines are what make a reader believe retry exists, so the unit reads as self-healing and is not; (2) systemd logs 'Directory /data-venus to mount over is not empty, mounting anyway' - the mountpoint has content underneath it, so a failed or unmounted state SHADOWS local files and a write there lands on whey's disk instead of venus, invisibly. lezama has since been provisioned with the .automount pattern (mount-on-access, so no boot-race dead state, reconnect+keepalive options kept) over an empty mountpoint, so the fleet is now inverted: lezama has the correct shape, whey has the broken one. DECISION NEEDED: adopt lezama's .automount shape on whey, or route the redesign to whoever owns whey's mount units. bin-whey-cc has not touched the unit beyond the manual start - no changes made pending this call. Whichever way it goes, the non-empty mountpoint must be resolved separately from the unit shape: .automount over a populated directory keeps the shadowing hazard.