I go with c
You only need provide wake turbulence separation within the path of the C402, and that is the absolute answer in Canada as far as I've been taught.
Despite the fact that wake does fall through the air, hence the 1,000ft when within 6nm ahead of the path of the light, it won't fall on faster on the path of the light if descent is delayed until through that path.
We regularly use such separation here against VFR traffic, against whom the radar separation is target resolution (or 1nm which amounts to the same thing). Cross over the top of the VFR a/c at close to 90 degrees, and descend as soon as target resolution is achieved, so long as the VFR's path is predictable or steady.