arm_neon: align ARM ASM to 4 bytes to work-around the linker on iOS

This matches a similar problem on x264: http://mailman.videolan.org/pipermail/x264-devel/2012-January/009073.html

modules/arm_neon/i420_rgb.S

@@ -74,7 +74,7 @@ coefficients:
     .short    4992
     .short  -18432
 
-	.align
+	.align 2
 	.global i420_rgb_neon
 	.type	i420_rgb_neon, %function
 i420_rgb_neon:

modules/arm_neon/i420_yuyv.S

@@ -35,7 +35,7 @@
 #define COUNT	ip
 #define OPITCH	lr
 
-	.align
+	.align 2
 	.global i420_yuyv_neon
 	.type	i420_yuyv_neon, %function
 i420_yuyv_neon:

modules/arm_neon/i422_yuyv.S

@@ -31,7 +31,7 @@
 #define COUNT	ip
 #define YPAD	lr
 
-	.align
+	.align 2
 	.global i422_yuyv_neon
 	.type	i422_yuyv_neon, %function
 i422_yuyv_neon:

modules/arm_neon/nv12_rgb.S

@@ -74,7 +74,7 @@ coefficients:
     .short    4992
     .short  -18432
 
-	.align
+	.align 2
 	.global nv12_rgb_neon
 	.type	nv12_rgb_neon, %function
 nv12_rgb_neon:

modules/arm_neon/nv21_rgb.S

@@ -74,7 +74,7 @@ coefficients:
     .short    4992
     .short  -18432
 
-	.align
+	.align 2
 	.global nv21_rgb_neon
 	.type	nv21_rgb_neon, %function
 nv21_rgb_neon:

modules/arm_neon/s32_s16.S

@@ -27,7 +27,7 @@
 #define	BUF	r3
 #define HALF	ip
 
-	.align
+	.align 2
 	.global s32_s16_neon
 	.type	s32_s16_neon, %function
 	@ Converts fixed-point 32-bits to signed 16-bits

modules/arm_neon/yuyv_i422.S

@@ -31,7 +31,7 @@
 #define COUNT	ip
 #define YPAD	lr
 
-	.align
+	.align 2
 	.global yuyv_i422_neon
 	.type	yuyv_i422_neon, %function
 yuyv_i422_neon: