i386 SSE2 Huffman: Fix pointer arithmetic issue

Splitting the pointer arithmetic in GET_SYM() into a separate add and
sub instruction was an attempt to work around an error ("invalid operand
type") that occurred when assembling the file with NASM.  However, this
created a link error on macOS ("ld: illegal text-relocation to
'_jconst_huff_encode_one_block' in
simd/CMakeFiles/simd.dir/i386/jchuff-sse2.asm.o from
'_jsimd_huff_encode_one_block_sse2' in
simd/CMakeFiles/simd.dir/i386/jchuff-sse2.asm.o for architecture i386")
and also changed the alignment of the code in ways that might have
affected the previous benchmark results (which took a great deal of time
to obtain.)  Ultimately, the path of least resistance is just to
require NASM 2.13 or later.

BUILDING.md

@@ -12,10 +12,7 @@ Build Requirements
 
 - [NASM](http://www.nasm.us) or [YASM](http://yasm.tortall.net)
   (if building x86 or x86-64 SIMD extensions)
-  * If using NASM, 2.10 or later is required.
-  * If using NASM, 2.10 or later (except 2.11.08) is required for an x86-64 Mac
-    build (2.11.08 does not work properly with libjpeg-turbo's x86-64 SIMD code
-    when building macho64 objects.)
+  * If using NASM, 2.13 or later is required.
   * If using YASM, 1.2.0 or later is required.
   * If building on macOS, NASM or YASM can be obtained from
     [MacPorts](http://www.macports.org/) or [Homebrew](http://brew.sh/).

simd/i386/jchuff-sse2.asm

@@ -252,8 +252,7 @@ times 1 << 14 db 15
     call        %%.geteip
 %%.ref:
     %4
-    add         %1, %2
-    sub         %1, %%.ref
+    add         %1, %2 - %%.ref
     jmp         short %%.done
     align       32
 %%.geteip: