The biggest issue with shield reps is the way they boost eHP for armor tanks aswell.

When you're being armor repaired, chances are, you are already or very soon to be out of shields and into cooldown. This means that the amount of eHP gained through armor reppers before you die is generally bound by your armor buffer+passive repair. This makes armor reps most effective on armor tanks and rightly so.

Conversely, a shield repaired suit gets full benefit of these repairs even if it's actually an armor tank. The repper keeps providing more eHP by adding small amounts of extra shields for the whole time until the all armor is depleted and the suit lost. Thus, hybrid/shield reppers provide equal or increased benefit to hybrid/armor tanks which is exactly backwards to what I assume is their intented effect.

Concerning hybrid reps. It also has to be considered that the two tank paradigms are designed to work differently and their numbers are set accordingly. If you make shields and armor more mechanically similar, you might end up having to make them numerically more similar.

When one paradigm doesn't work that well where it should, the first step should be to tweak some numbers around to better define the strengths and weaknesses of both tank types before changing the way both function on a pretty fundamental level. Remember that shields would gain a function ("can be repaired") they weren't necessarily intended to have* and armor would get, essentially, another type of repper they partially/fully benefit from.

An alternative to shield reps which avoids some of these issues would be a shield resistance boosting module. This gives shield dedicated logibros a way to support shield tanks but without benefiting armor tanks too much along with it as it doesn't circumvent the shields innate dependency on recharge and cooldown the way reps would do.

*On infantry, anyways. The Caldari LLAV had infantry targeting shield reps but both are gone right now and no one knows if they get to keep this particular ability on their return.