I won't pretend that I understand the physics of this, but one possible explanation is in FAA Order 6750.16E - Siting Criteria for Instrument Landing Systems. Section 3 discusses terrain effects on the ILS signal, and it notes that approaches that are partially over water may have signal quality issues. One way to address that is using a lower glide angle, per this section on page 3-4 (emphasis mine):
Where the Fresnel zone is smooth but not consistent, as where the
approach path is partially over water, a change may be encountered in
the received signal as the coefficient of reflection changes abruptly.
This effect is unavoidable; however, it can be minimized by
establishing the lowest possible glide angle (and, therefore, the
lowest possible grazing angle) and/or locating the glide slope
antennas within the specified criteria, so that the change in the
received signal does not occur within a critical part of the approach.
I don't know exactly which procedure you're looking at, but Larnaca is a coastal airport and it looks like approaches to 04/22 are at least partially over water.
I have no idea if the same issue occurs with other types of terrain. Someone mentioned in comments that the ILS at KTUL (Tulsa, Oklahoma) has a 2.75° glideslope landing south, but 3° landing north. KTUL isn't on the coast and there are no large bodies of water directly north of it. Maybe the same signal issue occurs there for different reasons, or maybe there's a completely different reason.