That's one definition of the leakage inductance. It seems that for power electronics, most people think it's acceptable to use that "lumped" leakage inductance instead of considering primary and secondary leakage inductance separately. I think that's valid for topologies such as flyback, forward, etc, but maybe not for LLC (I don't know, I'm no expert on LLC). But in any case, the single leakage inductance model is not equivalent to a two-leakage model, it's just an approximation. I also believe that even a two-leakage model is not a complete model, since it assumes the leakage inductances are proportional by N^2 (which is not necessarily the case).