Misc. cleanup for av1_round_shift_array_sse4_1

Document the 16-byte alignment requirement of the `arr` parameter of
av1_round_shift_array_sse4_1().

Declare the `input` parameters of av1_round_shift_array_32_sse4_1() and
av1_round_shift_rect_array_32_sse4_1() as const.

Change-Id: I116c84a477fb80eec89f928abcc38fcaed886d9a
diff --git a/av1/common/x86/av1_txfm_sse4.c b/av1/common/x86/av1_txfm_sse4.c
index 65ccd19..1894efd 100644
--- a/av1/common/x86/av1_txfm_sse4.c
+++ b/av1/common/x86/av1_txfm_sse4.c
@@ -14,6 +14,7 @@
 #include "av1/common/av1_txfm.h"
 #include "av1/common/x86/av1_txfm_sse4.h"
 
+// This function assumes `arr` is 16-byte aligned.
 void av1_round_shift_array_sse4_1(int32_t *arr, int size, int bit) {
   __m128i *const vec = (__m128i *)arr;
   const int vec_size = size >> 2;
diff --git a/av1/common/x86/av1_txfm_sse4.h b/av1/common/x86/av1_txfm_sse4.h
index 6cad821..387dfd6 100644
--- a/av1/common/x86/av1_txfm_sse4.h
+++ b/av1/common/x86/av1_txfm_sse4.h
@@ -25,7 +25,7 @@
   return _mm_srai_epi32(tmp, bit);
 }
 
-static INLINE void av1_round_shift_array_32_sse4_1(__m128i *input,
+static INLINE void av1_round_shift_array_32_sse4_1(const __m128i *input,
                                                    __m128i *output,
                                                    const int size,
                                                    const int bit) {
@@ -42,7 +42,7 @@
   }
 }
 
-static INLINE void av1_round_shift_rect_array_32_sse4_1(__m128i *input,
+static INLINE void av1_round_shift_rect_array_32_sse4_1(const __m128i *input,
                                                         __m128i *output,
                                                         const int size,
                                                         const int bit,