Molecules crystallize by ordering themselves in a crystal lattice.  There must be no or very minimal impurity, as they will not fit into the crystal lattice and prevent the very molecules from entering the crystalline phase.  You must have seen apples or oranges in the grocery store that are on display packed together in an orderly and stable fashion in 3D. That's how molecules are when they are in a crystal lattice, only at a much smaller scale.  So yes, there must be interactions - within the molecule itself, within the asymmetric unit, between molecules and between asymmetric units.

It is not necessary to have directional interactions like hydrogen bonds or pi-stacks. 

Many crystal structures are held together by dispersion-forces, which are non-directional long-range forces. Remember that even the rare gases and alkanes crystallise at low enough temperatures. 

Remember that Mercury is a visualiser that is useful for looking at the arrangement of molecules, but it has only very basic functionality for looking at the interactions between molecules: Mercury bases this mostly on finding atom pairs that are closer than some cutoff distance, such as the sum of van der Waals radii. It is important to realise that interactions do not switch off at this distance. Most of the interactions between molecules are much longer-ranged than this - dispersion and particularly electrostatic interactions are still significant well beyond van der Waals radii. If you want to assess the interactions that are holding the crystal structure together, you need to perform some calculations using force fields or an electronic structure method.