avcodec/simple_idct: use predefined M_SQRT2

M_SQRT2 is defined in math.h, or in avutil/mathematics.h for
compatibility hacks. This uses this value instead of a floating literal.

Fixed point values produced by C_FIX(), R_FIX() remain identical.
Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com>

libavcodec/simple_idct.c

@@ -132,7 +132,7 @@ void ff_simple_idct248_put(uint8_t *dest, int line_size, int16_t *block)
 #undef C1
 #undef C2
 #define CN_SHIFT 12
-#define C_FIX(x) ((int)((x) * 1.414213562 * (1 << CN_SHIFT) + 0.5))
+#define C_FIX(x) ((int)((x) * M_SQRT2 * (1 << CN_SHIFT) + 0.5))
 #define C1 C_FIX(0.6532814824)
 #define C2 C_FIX(0.2705980501)
 #define C3 C_FIX(0.5)
@@ -159,7 +159,7 @@ static inline void idct4col_add(uint8_t *dest, int line_size, const int16_t *col
 }
 
 #define RN_SHIFT 15
-#define R_FIX(x) ((int)((x) * 1.414213562 * (1 << RN_SHIFT) + 0.5))
+#define R_FIX(x) ((int)((x) * M_SQRT2 * (1 << RN_SHIFT) + 0.5))
 #define R1 R_FIX(0.6532814824)
 #define R2 R_FIX(0.2705980501)
 #define R3 R_FIX(0.5)