Skip some Math.atan2() tests.

Linux `atan2(3)` man page says:

```
If  y is positive infinity (negative infinity) and x is positive infinity,
+pi/4 (-pi/4) is re‐ turned.
```

But on Microsoft VC/MinGW, `atan2()` returns `NaN` if either of
arguments is infinite. So we skip those tests on the platforms.

mrbgems/mruby-math/test/math.rb

@@ -209,17 +209,18 @@ def assert_float_and_int(exp_ary, act_ary)
   assert_float(+Math::PI, Math.atan2(+0.0, -0.0))
   assert_float(-Math::PI, Math.atan2(-0.0, -0.0))
 
+  assert_float(0, Math.atan2(0, 1))
+  assert_float(Math::PI / 4, Math.atan2(1, 1))
+  assert_float(Math::PI / 2, Math.atan2(1, 0))
+
   inf = Float::INFINITY
+  skip "Math.atan2() return NaN" if Math.atan2(+inf, -inf).nan?
   expected = 3.0 * Math::PI / 4.0
   assert_float(+expected, Math.atan2(+inf, -inf))
   assert_float(-expected, Math.atan2(-inf, -inf))
   expected = Math::PI / 4.0
   assert_float(+expected, Math.atan2(+inf, +inf))
   assert_float(-expected, Math.atan2(-inf, +inf))
-
-  assert_float(0, Math.atan2(0, 1))
-  assert_float(Math::PI / 4, Math.atan2(1, 1))
-  assert_float(Math::PI / 2, Math.atan2(1, 0))
 end
 
 assert('Math.ldexp') do