Checkpoint.  Manipulated things so that string literals are always
unicode, and a few other compensating changes, e.g. str <- unicode,
chr <- unichr, and repr() of a unicode string no longer starts
with 'u'.  Lots of unit tests are broken, but some basic things
work, in particular distutils works so the extensions can be built,
and test_builtin.py works.
diff --git a/Python/ceval.c b/Python/ceval.c
index fdee13d..00fc050 100644
--- a/Python/ceval.c
+++ b/Python/ceval.c
@@ -2633,7 +2633,7 @@
 			PyObject *keyword = kws[2*i];
 			PyObject *value = kws[2*i + 1];
 			int j;
-			if (keyword == NULL || !PyString_Check(keyword)) {
+			if (keyword == NULL || !(PyString_Check(keyword) || PyUnicode_Check(keyword))) {
 				PyErr_Format(PyExc_TypeError,
 				    "%.200s() keywords must be strings",
 				    PyString_AsString(co->co_name));