I have been reluctant to respond, because my design for a high-current control using IRF3205's is very old and crude. I built it to control a 12V-winch that was made from a 6V Ford, long-shaft started motor (used in trucks, tractors, and old cars). Those motors were beasts, and at 12V draw well in excess of 100A, probably in excess of 200A when loaded. We loaded them almost to stall.
What you have been told is absolutely correct. An industrial mosfet will probably work better. But they cost a lot more. Second, read footnote #5 in the datasheet. It explains that the junction limitation for the 3205 may be 100 A, but the package limitation is only 75A. I think that is optimistic, for a continuous load. Heat is the issue. You have to keep the solder from melting.
Anyway, here are some pictures of what I did. I used the LT1158 to drive the top and bottom mosfets in a half bridge. You would need another for a full bridge. But since, the starter motor is not reversible, that was a non-issue for me. I paralleled 5 IRF3205's for each leg of the half-H controller. In the picture, you will see the typical gate resistors and a larger value resistor from the gates to ground. That was necessary as a "fix" so the gates would shut down. It may not be necessary in your application with the gate driver you chose. The conductors were made from 0.025 " copper plate and mosfets were soldered directly to it to aid in heat dissipation.
While I am sure this design will handle well over 100A for short periods, I am much less sure how it would work for continuous use at that current. At a minimum, you may need better heat sinks. Note, the connector shown on the schematic is only for control of the mosfets. Power to the source and drains was by fat cables.
John