WMO doodad sets use wrong index when loading

[fallenoak]

Turns out MODR indices should be taken against MODD, and not against MODD-intersected-by-MODS.

We should adjust doodad loading in WMOHandler to subtract MODS.start from doodadIndex prior to pulling the entry out of the filtered doodad set.

[timkurvers]

I'm assuming the current implementation should result in some funky stuff by loading the wrong doodads? Do we have any examples? Would love to see that 😄

[fallenoak]

It's less that wrong doodads are loaded (although I'm sure if we looked closely, we'd see some wrong doodads), and more that doodads wind up associated with wrong WMO groups. I figured out this was happening when I implemented portal culling -> doodad visibility and wound up with very weird doodads disappearing or reappearing.