Does the term "diatomic ideal gas" make any sense?

Of course it does.

It helps a little bit to compare the ideal gas to a model that does take note of the size of the molecules and the forces the exert on one another. The van der Waals gas has explicit parameters for both behaviors. Compare the equations of state for these two models \begin{align} Pv &= k_B T \tag{ideal gas} \\ \left(P + \frac{a}{v^2}\right)(v - b) &= k_B T \tag{van der Waals gas} \;, \end{align} where $a$ represent the net attraction between molecules, $b$ represent the volume actually occupied by a molecule, and I have written $v = V/N$ for the average volume available to any given molecule (so that $b/v$ is the fraction of the volume occupied by molecules a thing that is assumed to vanish in the ideal case).

This more complicated model exhibits the ability to condense into a liquid, a behavior that the ideal gas model does not duplicate. However, at low density and high temperature the van der Waals gas has the same heat capacity as the ideal gas.

On the other hand a non-monatomic ideal gas simple gets the addition of rotational and vibrational internal modes, and continues to use the same equation of state, though the new assumptions affect the heat capacity at all temperatures, but doesn't allow the system to exhibit condensation. If you take into account the effect of quantum mechanics on these internal modes the heat capacity exhibits steps in data on real gases that the monoatomic ideal gas does not faithfully report.

So, the van der Waals gas and introducing internal molecular modes represent two different way to introduce more complicated physics to the simplest gas model.

Either way we are extending a useful but incomplete model to allow it to capture more observed behavior than the naive version.