I've not worked on Tomato so am unfamiliar with genome browsers for this, though you would ideally want to target the first common exon across splice variants. The idea being that your gRNA would then induce an early truncation which should result in knockout for the alternate transcripts.

I would instead suggest targeting conserved coding domains of your gene of interest. Align some relative species' aa products, and the regions with hghest and most far-reaching conservation are likely the ones most sensitve to perturbation