Now here's the thing, I actually know why there is a difference.
When I bought a cheap co2 laser cutter, I did a lot of research into what type of water to use as the coolant for the discharge tube.

And water is never just water.
So what you gave there from your dehumidifier us actual de-ionised water.

And it has several peculiar properties that tap water does not have.

1) it has extremely low conductivity.
Turns out it's the dissolved minerals in ordinary water that give it it's excellent conductivity.
2) de ionised water will dissolve things that ionised water won't. It's considered a super solvent.
3) water is loosely held together by weak covalent bonds, when you freeze it the bonds between the water molecules get weaker and further apart, which is why water expands when frozen.  All the usual minerals and metals in a 'normal' water solution will help make the internal molecular bonds that little bit stronger.

So, taking that in mind.

It probably  sticks to the moulds more because they are not pure silicon and over the years of usage will have accumulated a residual coating.
The deionised water, is probably dissolving any impurities and creating micro pockets in the silicon which give it a better grip the standard water.

Likewise the weaker internal bonds in the frozen deionised water, will cause it to break and shatter more easily.

It's amazing what rabbit holes you end up going down when buying something as seemingly unrelated as a laser cutter

The other thing, which you can check if you have digital calipers.

Is that the deionised ice balls should be ever so slightly larger, than the tap water balls.

Another reason for them to be slightly tighter in the moulds.

It's a good experiment