Things are more complicated than that: orthogonality can achieve zero coupling (like other less well-known angle combinations), but the condition is not sufficient: symetry is also required.
In the case of the offset orthogonal coils, the field lines look like this:
If you place the second coil as in case 2, a good proportion of the lines of both coils will be almost perfectly aligned (on the right, where the arrows are drawn), resulting in a good coupling.
In addition, the magnetic field decreases as the inverse cube of the distance, meaning a slightly shorter distance will result in a much higher coupling (at least that's true for the far-field region, but let's not complicate the problem more than necessary).
To achieve cancellation, the second coil would need to be centered wrt. the first.
Of course, having parallel coils is far from ideal, but in this case it will probably be better than offset orthogonal coils.
Note that the inductance will increase only if the phase is correct, otherwise the effect will be a reduction. In simple cases, it is easy to guess when the phase is concordant, but in more complex situations it is prudent to measure both cases: if there is no difference, the coupling is zero, but any difference means a coupling is present
Even more unexpected, parallel coils can achieve zero coupling too: if in case 1 the lower coil is translated to the left, there will be a spot where the coupling is null. However, this will negatively affect the coupling with the air coil because the symmetry will be lost