...The only way a round gets away from a berm is to let loose a round over the berm...
Not true. And I have seen...
...Tracer rounds that struck the berm were seen going airborne and travelling out to sea where the plane had previously been flying. Seems once the bullet strikes the ground there is no way to know if it will bury itself or skip off into the atmosphere and continue down range...
...as well. Berms are built to be soft material to absorb the impact and reduce blowback striking other materials like previous bullets, and — as much as practical — for bullets to strike as close to perpendicular as you can get it. Rocks and even trees can cause bullets to deflect their trajectory, as much as 100° (sideways or up, and slightly back at you, yes) with a significant component of their velocity. Side berms are critical for safety here and/or the reason a WHOLE firing line must always be shut down; strikes on things like target frames totally send buzzing bullets flying over heads on adjacent ranges.
Shallow strikes onto the ground can retain almost all of the velocity and skip or "roll" along the ground, even over obstacles such as a berm downrange from the initial impact point. This is also known and taught to at least some soldiers and Marines: don't stand /right/ against a wall, in an urban fight for example, as a richochet can travel along the wall, but even back 6" and you are less likely to be struck by those.
The ranges I shoot on almost entirely have significant dead space beyond all berms (and all shots face outside the range, no ranges beyond other ranges) with wooded floodplains extending several hundred additional yards that are club property and no one is allowed to go there even though behind a berm and well below the firing line level.
To the OP's issue, .22 LR is — so I have been told — rather more dangerous than many other projectiles because it is unjacketed soft lead so much more likely do deform instead of fragmenting, and relatively stubby so when it tumbles looses relatively less aerodynamic performance than a spitzer bullet will.
I cannot immediately find direct measured evidence of max richochet ranges for range-design for .22LR specifically, but there are many stories of them causing injury or death at long ranges:
There is a famous case of an incident and subsequent manslaughter through negligence prosecution that took place in this country last century - can't find a link. The shooter fired at a bird or squirrel or something in a tree, missed and the bullet trajectory took it over the summit of a rise or low hill that the shooter had assumed would provide a backstop. At some extreme range, over 1,100 yards, it plunged into the eye socket of a man who was picknicking with his family on the reverse slopes and he'd lain down face up to sunbathe. By very bad luck, the bullet trajectory took it straight through an eye into the brain. Expert ballistic forensic evidence in the trial said it terminal energy was only 10 or 11 ft/lb, but hit in the wrong place and you can still do major harm.
Experiments might be worth it, with the specific range. Get some .22 tracer (and do this in not dry conditions, set a fire watch to sit there, ideally with a thermal scope, for some hours afterwards, etc) with lots of observers, at different angles so they can see for sure where the skips go, and fire a number of rounds on a cloudy day or at dusk, see where they go on the actual ground you have.
If the range or safe beyond-range area is deep enough then intermediate-range skips may be safe in the sense they all land on the range property anyway. This could prove it out for anyone with eyes. Or, prove to you that it does pose a risk and present other data so you can work around the real world threats for course design in future.
maps.app.goo.gl 
