Take for example the Jan Allaerts MC2, generally considered to be one of the top MC’s with unbelievable specs (how about 70dB channel separation?). It has a coil impedance of 32 ohms and according to the specs the load impedance needs to be set at exactly 845 ohms. Not 800 or even 850, but 845. Why this precision when the result is system dependent and that system is unknown?
If I remember this correctly, Jan Allaerts developed his cartridge using an Isenberg phono stage which happened to use an 845 Ohm loading. So it may be the other way round: The cartridge is designed to work at 845Ohm form the start - everything else is moving away from how it was designed and measured.
The spec sheet of Lyra Atlas is even more strange: recommended loading is between 104 and 887 ohms. A wide range as usual, but why these very specific numbers at its limits? Perhaps jcarr can explain, as I assume there must be a logical explanation for this.
The range depends on the capacitance of the cable and optimizing the frequency response gives a corresponding value of the loading...those endpoints are calculated values from my understanding - likely based on a rounded capacitance value ;)