libretro stuff

[tom-seddon]

#376

• Include (or not) strlcpy based on HAVE_STRLCPY. Must be possible to get cmake to detect this
• Make the DiscInterface objects non-static. No massive reason not to if there'd be a useful use for it
• Follow through with more Messages -> LogSet stuff, as there's still a needless amount of it in the disc geometry stuff that's tying it to b2 (even though it's otherwise just a few tables and some helper routines)
• Can probably just give up on the no I/O in shared_lib rule (there's already a bunch of path stuff, and fopenUTF8...) - move some generic stuff from b2's load_save.h into shared_lib somewhere
• With I/O in shared_lib, and more LogSet stuff, can move some more useful stuff to beeb_lib: DirectDiscImage, at least, and surely MemoryDiscImage too
• Do something slightly better about log_printer_stdout. Could do with some kind of thing for routing stdout/stderr via a callback... or maybe it's always a callback, and the default is stdout/stderr, and you can change it at runtime. Goal would be to facilitate interop with libretro's output callback

[tom-seddon]

MemoryDiscImage does a bunch of zip file stuff (given a zip file containing a signle disk image, it will load it) - which will need a bit of sorting out. I think the bulk of it can still move into beeb_lib and the zip file stuff kept separate in b2.

[tom-seddon]

3c3797b does the strlcpy thing. I called the define SYSTEM_HAVE_STRLCPY, not wanting to make a mess by interfering with the same defines in libcurl and SDL.

[tom-seddon]

Don't know exactly what to do about the log printer. By the time the text actually gets printed (which is where the libretro log callbacks would step in), all the formatting has been done and any info about the provenance has been lost.