Handle malloc errors properly

[libass]

Handle malloc errors properly

Yeah, I also planned to do something about that. But for now it seems to be better to wait until all the other changes make it in.

If we do this, we should also treat integer overflows correctly.

E.g. I see stuff like: realloc(track->events, sizeof(ASS_Event) * track->max_events)

If the multiplication overflows, the allocation will be smaller than expected (but still "succeed"), and you'll have a problem.

[rcombs]

Yup.

Reminder that we really should do this (tm).

OK, here's a suggestion:

• replace all malloc() calls with a ass_xmalloc() macro, which abort()s on malloc failure
• code that actually checks for the malloc() result can still use malloc() proper
• run a fuzz tests that make malloc() return NULL randomly to verify that the malloc checks actually work
• add some helper macros to allocate arrays with avoiding integer overflows