fix irep float dump format string for MRB_USE_FLOAT

IEC 60559 single format has 6 to 9 significant decimal digits precision.

However the printf conversion specifier e (and E, of course) already
writes 1 digit - the one before the decimal point - and precision
specifies the number of digits to write after the decimal point.

src/dump.c

@@ -16,7 +16,7 @@
 #define FLAG_BYTEORDER_NONATIVE 0
 
 #ifdef MRB_USE_FLOAT
-#define MRB_FLOAT_FMT "%.9e"
+#define MRB_FLOAT_FMT "%.8e"
 #else
 #define MRB_FLOAT_FMT "%.16e"
 #endif