I think I need to be clear about this because there are several different descriptions and I have kind-of said both.
As Questionpost says, the probability of finding the particle(s) of interest is so small outside the lab we can treat it as zero to a very high degree of accuracy.
However - it is also possible to rig up an experiment so that the probability of finding the particle outside our equipment, at least for a time, is zero. For instance, a photon created inside a laser cavity has zero chance of being created, by the experimental setup, anywhere else.
Chemical reactions should occur if atoms are in some way "unhappy" with their current relationship or there is room for some orbital to take an electron with opposite spin, and the EM force of the protons along with room of the electron would allow an atom to give, take, or share an electron.
Nice one - now try for a
scientific description in terms of quantum mechanics ... you want to express your answer in terms of wavefunctions, energy levels, and potential wells. Let's keep it simple and go for a reaction like: \text{H}+\text{H}\rightarrow\text{H}_2 ... how does that happen?
I suppose covenlent bonds wouldn't really be possible at large distances,
Why not? This reason is central to your understanding.
but theoretically if the EM force virtual particles go on indefinitely, why does it even die out by the square of the distance? Is it just literally how it spreads out over 3D space?
You are confusing two different things - in the standard model the EM force does not exist ... except as an interaction between a photon and an electron. The classical force with it's famous inverse square law is a macroscopic phenomenon which is only obeyed on average.
The probability that a particular charge feels a force from another particular charge is the sum over all paths of the amplitudes of all the possible ways a photon could find it.
This sum produces the inverse square relationship where there is
spherical symmetry in the system. You can understand it by imagining the source sending out photons equally in all directions. The probability of an interaction will be proportional to the number of photons that are detected by the target particle - which is proportional to the number of photons passing through each unit area at that distance ... the photon flux density. You should be able to see that the flux density decreases with increasing area. Which is inverse-square.
It is more complicated than that but that gives you a mental picture of how the geometric symmetry of space produces an inverse-square law. It's also a pretty standard picture - I'm surprised you have not come across it before.