[Nano-X] Speed up VGA 4-plane 16-color display output

[ghaerr]

Should speed up VGA display output by quite a bit on real hardware. After (re)writing an AS86 assembly VGA driver for the 8086-toolchain in ghaerr/8086-toolchain#46, I noticed quite a few speed improvements that could be made to the C driver currently used in Nano-X.

The enhancements to the C vgaplan4.c driver include:

• Use OUTW instead of OUTB for writing VGA control registers
• Use kernel arch/io.h outw macro instead of calling function to output bytes to VGA
• Use LDS BX instead of loading DS and BX separately in RMW_FP, OR_BYTE, etc routines
• Save DS in DX instead of pushing to stack in same routines

@tyama501, I would be interested to know whether this speeds Nano-X up much on PC-98. In particular, large rectangle fills or screen clears should be much faster. (I am working next on getting UNIX sockets working on FAT filesystem).

The screen driver could probably be sped up even more by moving to a GCC assembly language version. I haven't done this yet, as that entails rewriting (again) the new C86 vga-4bp.s driver from Intel AS86 format to AT&T GCC format.