Remove alignment from loops. Newer cpus will execute the loop efficiently…

Remove alignment from loops.  Newer cpus will execute the loop efficiently without alignment, and the extra nops would slow the initial iteration marginally if anything.
BUG=none
TESTED=try bots
R=harryjin@google.com

Review URL: https://webrtc-codereview.appspot.com/27199004

git-svn-id: http://libyuv.googlecode.com/svn/trunk@1167 16f28f9a-4ce2-e073-06de-1de4eb20be90

include/libyuv/row.h

@@ -501,7 +501,7 @@ typedef uint8 ulvec8[32];
 #if defined(__native_client__)
 #define LABELALIGN ".p2align 5\n"
 #else
-#define LABELALIGN ".p2align 2\n"
+#define LABELALIGN
 #endif
 #if defined(__native_client__) && defined(__x86_64__)
 #define BUNDLELOCK ".bundle_lock\n"
@@ -541,7 +541,7 @@ typedef uint8 ulvec8[32];
     #opcode " $0x0,%%" #reg ",(%%r15,%%r14)\n" \
     BUNDLEUNLOCK
 #else  // defined(__native_client__) && defined(__x86_64__)
-#define BUNDLEALIGN "\n"
+#define BUNDLEALIGN
 #define MEMACCESS(base) "(%" #base ")"
 #define MEMACCESS2(offset, base) #offset "(%" #base ")"
 #define MEMLEA(offset, base) #offset "(%" #base ")"
@@ -567,9 +567,9 @@ typedef uint8 ulvec8[32];
 #if defined(__arm__) || defined(__aarch64__)
 #undef MEMACCESS
 #if defined(__native_client__)
-#define MEMACCESS(base) ".p2align   3\nbic %" #base ", #0xc0000000\n"
+#define MEMACCESS(base) ".p2align 3\nbic %" #base ", #0xc0000000\n"
 #else
-#define MEMACCESS(base) "\n"
+#define MEMACCESS(base)
 #endif
 #endif