They do have scientific uses.
The group I'm in, for example, uses them to probe Rydberg excitons in Cu2O (cuprous oxide) (link to on of our papers in Phys. Rev. Materials). The phosphor output of the lime LED is an almost perfect match to the spectral region of interest, providing greater spectral intensity than a halogen source, with less undesirable above-bandgap excitation.
I'm sure we're not the only scientific users, because Thorlabs, a supplier of laboratory optics package and sell them from stock.
In general, any wavelength range is useful for some niche applications. This has been obvious for years with lasers, but now there's such a wide range of high-power LEDs we can applications for those too. In related experiments we also use amber and blue LEDs from the same series.
By the way, as I hinted above, these aren't made up of two emitters, but are a blue LED with a phosphor on top, like white LEDs. The phosphor is rather slow, and early datasheets weren't as clear as the latest ones, so we only discovered this when we tried to pulse them at a timescale of tens of ns. It's actually rather obvious if you think about the width and shape (asymmetry) of the emission band, which doesn't match a typical emitter.