As @Jack says in his comment, 1 & 2 are the same. If you connected the resistor to the other side of the load in #2 it would dissipate 160W when on at 100%. That's not very practical.
If you use a DC voltage (option 3), you lose the isolation the opto-triac provides, which can make it a bit tricky to do firmware debugging (without blowing up your emulation pod, computer, etc.). If the triac or the connections to it go pins-up it could take out the circuit (think of an open between MT1 and ground).
It's better to drive the gate with a negative current. The triacs work better in those quadrants (you avoid the problematic quadrant IV - gate positive, MT2 negative). Some triacs are rated to work in that quadrant, some are not guaranteed but it is less sensitive in any case so it will not be as good (more EMI, due to later switching, for example).
Another option is to drive the triac gate through a pulse transformer. A single pulse or a pulse train can be used. The pulse has to be of sufficient length to allow the thyristor to turn on and the through current to build up sufficiently (despite whatever inductance is in the circuit) to reach the holding current of the triac. The pulse transformer can provide galvanic isolation, similar to the optotriac, and has certain advantages (more resistant to heat and aging).