[Bf-committers] prohibit use of long in DNA

[joe]

Our current SDNA system does not support longs correctly for the
different 64-bit data models in use today.  Under some compilers,
longs are 64-bit, while on others they're 32-bit.  At the moment, we
hardcode the length of longs as 4 bytes in several places, including
makesdna.

I don't think there's really any benefit to using longs in DNA.  In
fact, I've been told that supposedly they're not used anywhere in DNA
at the moment.  I think we should explicitly have maksdna reject
structs that make use of longs.  This way unsuspecting people won't
accidental use them, which could cause loading/saving .blends across
different systems to fail.

Joe