Use log2(b) instead of (log(b) / M_LN2)

Co-Authored-By: Nobuyoshi Nakada <nobu@ruby-lang.org>

math.c

@@ -538,6 +538,7 @@ math_log_split(VALUE x, size_t *numbits)
 # define log_intermediate log2
 #else
 # define log_intermediate log10
+double log2(double x);
 #endif
 
 VALUE
@@ -561,7 +562,7 @@ rb_math_log(int argc, const VALUE *argv)
             return DBL2NUM(-0.0);
         }
         d = log_intermediate(d) / log_intermediate(b);
-        d += (numbits - numbits_2) * M_LN2 / log(b);
+        d += (numbits - numbits_2) / log2(b);
     }
     else {
         /* check for pole error */