The quote you mention is for electric dipole transitions, which are not allowed for homonuclear diatomics. These transitions are for absorption or emission of one photon
You are correct to question absorption and re-emission; this is Raleigh scattering if the photon does not change wavelength (energy). The photon can change energy, though, in Raman scattering. The combined absorption and emssion process is generally very much weaker than allowed emission or absorption of one photon. Modern experiments generally observe Raleigh / Raman scattering with use of lasers, mostly in the visible I think, but no reason that it should not work in the IR as well. One final point is that in Raleight scattering the wavelength of light can change a little bit due to momentum transfer, but the internal energy of the molecule stays the same.
Final point is that some transitions that are not allowed as electric dipole transitions are allowed are electric quadrupole transitions or magnetic dipole transitions, for example. These processes are significantly weaker (less probable) than electric dipole transitions.