So…
You can do this. The best way I know how to do so is to snap the items together, then remove the overlapping path on one of the items.
Here’s the thing though: the amount of time and fiddliness of that process might make it irrelevant in most cases. Set the arrange to put them .5mm apart, and you’re losing almost nothing. In terms of cost, the labor cost of fiddling with your paths far outstrips the costs of almost any material I can think of… Let’s do a real world test.
On the left, zero space. On the right, .5mm space between.
The difference in materials is 9.18 square centimeters, or 1.42 square inches, which is roughly 1/100th of a square foot.
Take a typical material, like hardwood maple, at $6/square foot. You end up with a $0.06 material loss due to the spacing. I don’t know about your hourly rate, but 6 cents doesn’t buy much of my time. Lets be pessimistic and say you’re doing something more expensive, like that annoyingly expensive double sided mirror from Inventables. It clocks in at $46.49 for 12x24", so $23.24/square foot. That comes out to $0.23 in materials lost. Again, my hourly rate is way higher than that. It changes the equation if you save a bunch of laser time by eliminating duplicate paths, but it’s tougher to put a value on the idle time of laser job running.
So, unless you really like manipulating nodes, or are doing a massive production run where every second counts, or have very very tight space limits (which should be avoided, and is another topic entirely), there’s no real economic argument to be made in favor of joining shapes like that.