The gate driver doesn't do much for you in simulation, it's a propagation delay and that's about it; the particulars about UVLO, input threshold, output level, etc. shouldn't be too important. It's quite possible that's about all they're simulating in that model anyway, it doesn't look too involved.
Using a plain E source has no delay, but if delay is a small fraction of your overall loop phase margin (and, I would hope it is?), it won't have much impact; and you can simply avoid testing the other aspects as part of control loop modeling.
A part of modeling is wanting to build something as comprehensive as possible, for example so you can build the circuit, do all your simulations in one place with the complete model, then build it from exactly the same place. Well, you don't have that last luxury, as LTSpice isn't PCB design software, but to the extent you can import the schematic in something else, or at least the netlist, there would be that.
But a more refined level of modeling, or perhaps a substitute depending on particular goals of a given project -- or a necessity above a certain scale -- it can be broken up into smaller pieces. This may not be too important for a converter, which you can easily model whole; but if you're having trouble putting together the whole thing, you might still slice it up different ways. Namely: the inverter by itself, doesn't need to be tested in a closed loop; you can wire up a few VPULSEs for gate drive, with a dummy load, resistor or inductor or some combination like that, to get representative load current going. That's good for testing switching loop dynamics (rise/fall, ringing, overshoot). Conversely, the control loop doesn't care a smidge about the rise/fall/ringing, the inverter can basically be a single switch for all that matters, or even a dependent source.
There's also such a thing as an average mode model, which does away with the switching, and models the control loop very well; ripple can even be added back in; at least, potentially so. (However, to do this automatically, as it were, you'll need support from the sim engine itself, i.e. periodic steady state (PSS) analysis -- ngspice for example includes this.)
Tim