You do know that incandescent bulbs can pull about
10x the current on turn on?

You do know that having 700W of incandescents on even a 15A
circuit is very common and doesn't cause breakers tripping?

OK...Lets talk details.

I don't know the details of the inards of GFI breakers. Maybe you do.

Lets say the COLD turn on surge is 20 Amps for a short time and that alone is not enough to trip a 20 A breaker.

Lets say there is also 3 mA of leakage and the trip point for the GFI breaker is 5 mA so 3 mA alone is not enough to trip the breaker.

But what about both together? Maybe both together will trip.

I don't know if each trip point is totally seperate inside the breaker or if they somehow are added. Do you? (I'm not trying to be snooty)

I do know that a truely COLD bulb turn on draws a bigger surge then one where the bulbs have been pre warmed. It doesn't seem logical I agree, but I have seen it. You have to wait a good number of seconds for the filament to totally cool to get the full surge current.

It doesn't seem like this small diffence should be the OPs problem but at this point, who knows?

I agree with the suggestion to change to a non GFI breaker as a test to eliminate the leakage part of the question.

Mark