SF bug #513866: Float/long comparison anomaly.

When an integer is compared to a float now, the int isn't coerced to float.
This avoids spurious overflow exceptions and insane results.  This should
compute correct results, without raising spurious exceptions, in all cases
now -- although I expect that what happens when an int/long is compared to
a NaN is still a platform accident.

Note that we had potential problems here even with "short" ints, on boxes
where sizeof(long)==8.  There's #ifdef'ed code here to handle that, but
I can't test it as intended.  I tested it by changing the #ifdef to
trigger on my 32-bit box instead.

I suppose this is a bugfix candidate, but I won't backport it.  It's
long-winded (for speed) and messy (because the problem is messy).  Note
that this also depends on a previous 2.4 patch that introduced
_Py_SwappedOp[] as an extern.

Misc/NEWS

@@ -15,7 +15,14 @@ Core and builtins
 - The bytecode optimizer now folds tuples of constants into a single
   constant.
 
-- PyLong_AsUnsignedLong[Mask] now support int objects as well.
+- SF bug #513866:  Float/long comparison anomaly.  Prior to 2.4b1, when
+  an integer was compared to a float, the integer was coerced to a float.
+  That could yield spurious overflow errors (if the integer was very
+  large), and to anomalies such as
+  ``long(1e200)+1 == 1e200 == long(1e200)-1``.  Coercion to float is no
+  longer performed, and cases like ``long(1e200)-1 < 1e200``,
+  ``long(1e200)+1 > 1e200`` and ``(1 << 20000) > 1e200`` are computed
+  correctly now.
 
 Extension modules
 -----------------
@@ -72,6 +79,8 @@ Build
 C API
 -----
 
+- PyLong_AsUnsignedLong[Mask] now support int objects as well.
+
 - SF patch #998993: ``PyUnicode_DecodeUTF8Stateful`` and
   ``PyUnicode_DecodeUTF16Stateful`` have been added, which implement stateful
   decoding.