msvcrt: Use the erf()/erfc()/erff()/erfcf() implementation from the bundled musl library.

libs/musl/src/math/erf.c

@@ -269,5 +269,8 @@ double __cdecl erfc(double x)
 	if (ix < 0x403c0000) {  /* 0.84375 <= |x| < 28 */
 		return sign ? 2 - erfc2(ix,x) : erfc2(ix,x);
 	}
-	return sign ? 2 - 0x1p-1022 : 0x1p-1022*0x1p-1022;
+	if (sign)
+		return 2 - DBL_MIN;
+	errno = ERANGE;
+	return fp_barrier(DBL_MIN) * DBL_MIN;
 }

libs/musl/src/math/erff.c

@@ -179,5 +179,8 @@ float __cdecl erfcf(float x)
 	if (ix < 0x41e00000) {  /* |x| < 28 */
 		return sign ? 2 - erfc2(ix,x) : erfc2(ix,x);
 	}
-	return sign ? 2 - 0x1p-120f : 0x1p-120f*0x1p-120f;
+	if (sign)
+		return 2 - FLT_MIN;
+	errno = ERANGE;
+	return FLT_MIN * FLT_MIN;
 }