Siphash-2-4 is now our hash in nearly all cases.

I've made an exception for cases where I'm sure that users can't
influence the inputs.  This is likely to cause a slowdown somewhere,
but it's safer to siphash everything and *then* look for cases to
optimize.

This patch doesn't actually get us any _benefit_ from siphash yet,
since we don't really randomize the key at any point.

src/ext/siphash.h

@@ -1,5 +1,6 @@
 #ifndef SIPHASH_H
 #define SIPHASH_H
+
 struct sipkey {
   uint64_t k0;
   uint64_t k1;