I think what you are forgetting is that a given Feshbach resonance is between two atoms in a certain scattering channel. This means, in particular, that it only occurs for one partial wave. As far as I know, all observed Feshbach resonances have been either s-wave or p-wave, but presumably they also exist for higher partial waves. Anyway, the point is that for a room-temperature gas atoms interact with many of these partial waves, and the impact of any particular Feshbach resonance would be small.
The threshold energy for a partial wave with angular momentum $l$ goes as
$E_{th}(l)\sim A (l(l+1))^{3/2}$,
where $A$ depends on the inter-atomic potential and masses and might range from 10 $\mu $K - 1 mK. So, at the K scale and above, it would presumably be difficult to see the effects of any one collision channel. However, I do not know if anyone has tried this experiment, or if it would be feasible with some very careful measurement.
There are probably other considerations too. For example, remember that a typical ultracold atomic gas has something like $10^{-6}$ the density of air, which is important for two-body collisions to dominate the physics.