Having thought about this more, I think it suits my sand table more than the 'classic' homing approach, so I'm going to adopt it. Thanks @R006 for highlighting it. (On teh other hand, 99% of the time I don't need the table homed on this axis - the pattern will just face a different direction, and it's truly unlimited continuous rotation mechanically.)
My previous code snippet misbehaves if the endstop is already triggered when first called and the 'touch more' movement drives it out of triggered, so I've actually adopted:
G91 ; relative moves M574 Y1 S1 P"ystop" ; proper endstop polarity if sensors.endstops[1].triggered = false G1 H1 Y-720 F1800 ; do homing move @1800 (5 rpm, 12 secs per rev) G1 H2 Y-2 F1800 ; move slightly further - endstop zone is approx 6 wide M574 Y1 S1 P"!ystop" ; reverse polarity G1 H1 Y10 F360 ; home to release at 1 rpm G92 Y0 ; set zero M574 Y1 S1 P"ystop" ; reset endstop polarity G90 ; absolute moves