Trouble with mounting holes and connectivity to shapes

We used to use simple plated through holes for screw mounting, but a DFM review caused to us pre-emptively remove through hole barrels from under screw heads for fear of crushing the barrels. Our new mounting holes have 4 through holes outside of the screw head. There is an exposed copper ring, a shape, with both top and bottom Clines poking out of the copper ring to the through holes. To make life easy, we use a single pin with 3 vias (pin is in lower left of these screenshots), so engineers only have to make one connection on the schematic.

Although connectivity can't be assigned in the symbol, we noticed that dropping down the symbol in an unused area of the PCB assigned all copper and the vias to the net of the pin (gnd_sig2). Notice no DRC errors and all copper elements are properly connected together. [1.png]

I'll first plop this symbol onto a board with a shape on the bottom layer (green) assigned to gnd_sig1. As you can see, the vias decide to forget about the pin's net assignment and they get connected to the shape. The Clines touching the vias switch to gnd_sig1 as well, along with the copper ring on the bottom (green). You can see that the copper ring has voided around the pin's Cline, which is still connected to gnd_sig2. The copper ring on the top (red) also stays connected to gnd_sig2, so it doesn't void, but it creates DRC errors with the Clines on the vias which are now gnd_sig1. If I move the symbol off the shape, the vias report no net connectivity. [2.png, below]

Next, I'll drop the same symbol onto the same shape, but moved to an inner etch layer (blue, also set as TYPE=CONDUCTOR) and still assigned to gnd_sig1. This time, the copper ring on the bottom (green) doesn't void but the vias (and Clines) again attach to the shape while the pin is left unconnected as it's still assigned to gnd_sig2. DRC errors are again created. [3.png, below]

Finally, I'll drop the symbol down all alone and then extend the same inner layer etch shape (blue) to cover the symbol. You can see the vias aren't connected to the shape as they, and all other copper elements, retain gnd_sig2 attachment. This is how I'd expect it to always work. [4.png, below]

So, I've noticed different behavior if the shape exists before placing the symbol, than if the shape is added after the symbol is placed. Ideally, the net would be "fixed" to all elements with connectivity to the pin, in other words the only element with net connectivity (the pin) should drive connectivity to the rest of the symbol's elements, but that doesn't seem to be the case.

Adding a Cline arc that touches the Cline "spokes" would probably fix this, or changing the copper ring to a thick Cline arc, but before I change anything I'd like to get your feedback. Are our mounting hole symbols poorly designed for other reasons? Can we do something better? Is there a better fix than what I just proposed? Are we not properly using the SPB tools, which causes this issue? What do you guys do?

Thanks!

1.png