First, and as you stated, the photon energy links to the bandgap, therefore depending on the value of the bandgap (so the material), photons with wavelengths outside of the visible spectrum are possible.
- By indirect bandgap, it is understood that the maximum of the valence band and the minimum of the conduction band are not found at the same wavevector. Therefore, for an electron-hole pair recombination to happen, this $\Delta k$ needs to be accounted for. Usually this is done via the emission of a phonon. This makes the entire recombination process less efficient for indirect band gap semiconductors compared to direct ones as both an electron/phonon and an electron/hole interaction are needed (same electron involved in the entire process).
- Nonradiative recombination is also possible, with the emission of a phonon for example (may involve defects level in the bandgap) or by supplying the energy to a third carrier (Auger recombination).
- At first order, given that the electrons and holes injected respectively from the n and p side of the junction both obey the Fermi-Dirac statistics, a true monochromatic emission is not possible (not considering the ideal case of 0K which would in any case raise other issues for the device operation).
Concerning MOSFETs, note that they do no involve electron-hole pair recombination as they primary conduction mechanism. They are based on a single majority carrier, NMOS for electron conduction and PMOS for hole conduction. Although we cannot rule out recombination (or even generation) as a whole, it is rather a source of noise, so no your MOSFETs would not look like a glowing board of LEDs, at least if we are not talking about thermal radiation which does not find its origin in free carrier recombination.