Victor Møller Pedersen

Needs to be fixed already...

Updated (1.6.2)

Updated. The issue is oddly specific to tree growth, although I reckon the sensor is ultimately the problem since it is not updating correctly. I don't know if it's related to some kind of update cycle and some results just randomly got the right timing...

It happens randomly in all conditions, be it flowing or still water, no matter how it is dispensed, block or not. It just happens sometimes.

You're right. The half hearts confuse me. Mistake on my part.
When I say 19½ I mean just 19, while 19 would be 18. I'll change it.

Also note that you may not be able to reproduce by jumping off a cliff yourself. I tested it with Creepers in a controlled environment. Creepers have 20 health, and by falling the correct distance, should be left with 1 health. In order to verify this, I simply attack them barehanded afterwards, which is known to do 1 point of damage. Yet I had to hit them twice if they survived, which means they were left with 18 health.

Not limited to the top of the stairs either. I discovered it when trying to convey items under stairs with water currents. Described in detail in duplicate: MC-3572 Items and stairs buggy collision

Finally, we have Screenshot 4. This is a test to show that it isn't exclusive to water.

Item 1 was thrown directly at the middle opening of the stairs in an attempt to get it through. The result was that it was pushed to the side, towards Item 2, which was thrown at a normal block.
If Item 2 is thrown first, Item 1 will land exactly on top, essentially stacking the two. You cannot tell there is more than one item, as the animation will also sync perfectly.
Note that if Item 1 hits the top of the stairs, it will act as if hitting a regular block like Item 2 does.
Item 3 was thrown like Item 1 but at a slab (aiming for the opening), and passed under without problems.

In Screenshot 3, the item is enclosed. In this setup, the item will simply bounce around indefinitely, often several blocks away if not for the water current. Rarely, the item will finally pass under the stairs, but this can take several seconds.

After bouncing around for a while in Screenshot 1, the item is eventually pushed to the side in Screenshot 2.

Screenshot 1 shows a simple setup that uses water to attempt pushing an item under some stairs. In this case, one wall is exposed. This is important due to what happens in Screenshot 2.

I've tested it, and the issue was not present in previous versions. Updated.

This should be considered a bug. In general, tripwire shouldn't be destroyed when blocks underneath are changed. It kills a lot of the potential and depth of the system. By changing this, we can finally have a detection method on the ground that still works without said ground. For instance, pulling away the floor without the tripwire breaking.

Also I want to point out, you can have a grounded tripwire without a block under one of the hooks. I'm sure it relates to the whole N/S/E/W thing, as only a specific end will work.
Hell, while I'm at it, I think hooks themselves should trigger the mechanism too. But the issue at hand is as OP stated.