Improve FatFs speed by merging wf-fatfs changes

[rasky]

https://github.com/WonderfulToolchain/wf-fatfs is a FatFs fork which contains some interesting speedup. For instance:

#define FF_WF_FAST_CONTIGUOUS_READ  0
#define FF_WF_FAST_CONTIGUOUS_WRITE 0
/* FF_WF_FAST_CONTIGUOUS_* controls whether or not contiguous reads or writes
/  of more than 1 cluster (>4-32KB) are optimized to use large disk_read()
/  and disk_write() calls. This can be useful on platforms where the cost of
/  initializing a sector read/write is large.
/
/  0: Do not optimize this scenario.
/  1: Optimize this scenario.
*/

This should be able to speed up ROM loading (as performed by flashcart menus) by quite a bit, by issuing even just a single read operation.

We should review the change in wf-fatfs and incorporate them if applicable to N64.

[asiekierka]

How do you feel about submodules? That's how at least the projects I worked with integrated wf-fatfs into their own trees.