NWScript object management

[seedhartha]

This PR makes NWScript variables store object ids, not object pointers. It introduces two new classes: ObjectReference, which holds the object id, and ObjectManager, which acts as the global registry for objects. This PR also implements destroyObject function in KotOR.

[DrMcCoy]

There's still use-after-free crashes here: https://gist.github.com/DrMcCoy/0fd8777a1a179bd921de3192895f24d0

The reason seems to be that the _owner and _triggerer member variables in NCSFile are still plain pointers. When assigning those to stack variables, the ObjectReference tries to get their ID, crashing.

[DrMcCoy]

Hmm, Placeable::load() reads the _id out of the GFF, though. There's probably no reason it has to be this _id, though, so you can probably add a new variable, say _placeableID or something, and assign the value from the GFF there.

[DrMcCoy]

For areas, the _id gets overwritten in Area::loadEnvironment(), with a value from the GFF. You'll need to add a commit adding a, for example _envID value to Area and change the assignment in loadEnvironment to use write into that value instead.

Same applies to Dragon Age II as well.

[DrMcCoy]

Otherwise, I like this PR. Looks to me like it goes in the right direction :)

[seedhartha]

Fixed the above issues.

[DrMcCoy]

Sorry, got another use-after-free here: https://gist.github.com/DrMcCoy/549d6f2ed5ea7532539ee148a02f75a7

The issue here is that Module::_delayedActions contains raw Object pointers (ActionQueue, which is a std::multiset, struct Action members owner and triggerer)

[seedhartha]

Done.

[DrMcCoy]

I can't seem to trigger any crashes now, yes. :)

However, could you apply bc7047f to the other engines? All except for Sonic have that struct Action with a raw Object pointer in module.h.

[seedhartha]

Modified actions/events in other engines as well.

[DrMcCoy]

Merged as 08d31ac...61cd69b, thanks! :)

[seedhartha]

Have you intentionally not included the last DA commits or is something wrong with GitHub today?

[DrMcCoy]

Huh, yeah, I somehow missed them.

GitHub is weird today, a bit slow with the updates. But this might be completely on me, I think I forgot to update the remote before merging. I'll add them, sorry

[DrMcCoy]

There, added them as 61cd69b...6c7e9c6. And made sure the commits I merged are the correct ones and nothing went missing.