overlay: fix regression in pickup display

[lahm86]

Resolves #213.

This was introduced in a7604d3. I am assuming there is something not quite adding up in the underwater pickup collision (yet to be decompiled). As we retain int16_t for the time being in ITEM_INFO, this solution fixes the problem, although it's less than ideal. LMK if you think there is a better stopgap.

[rr-]

I think the only place that needs updating is the function declaration and definition, the other places will promote the type automatically.

TBH I don't understand why this crashes. I get that the size of GAME_OBJECT_ID is 4 bytes whereas the size of int16_t is 2, but x86 should align each parameter in function calls to 4 bytes anyway: https://stackoverflow.com/questions/30679702/why-function-parameter-occupy-at-least-4-bytes-stack-on-x86

[lahm86]

I think the only place that needs updating is the function declaration and definition, the other places will promote the type automatically.

TBH I don't understand why this crashes. I get that the size of GAME_OBJECT_ID is 4 bytes whereas the size of int16_t is 2, but x86 should align each parameter in function calls to 4 bytes anyway: https://stackoverflow.com/questions/30679702/why-function-parameter-occupy-at-least-4-bytes-stack-on-x86

It makes no sense to me either. I guess once we have pickup collision decompiled we can go back to the way it was. I can't imagine it's called from anywhere else, other than the gameflow and on pickup, but we'll see. It's so strange that there is no issue above water, even if you drain an underwater room with a pickup.
Code tidied up, so redundant casts removed in gameflow.c, All pickups seem good, and gameflow displays are still working e.g. secret rewards.