Implement RFC#28: Add PartialOrd::partial_cmp

I ended up altering the semantics of Json's PartialOrd implementation.
It used to be the case that Null < Null, but I can't think of any reason
for an ordering other than the default one so I just switched it over to
using the derived implementation.

This also fixes broken `PartialOrd` implementations for `Vec` and
`TreeMap`.

RFC: 0028-partial-cmp

src/test/compile-fail/deriving-span-PartialOrd-enum-struct-variant.rs

@@ -27,6 +27,7 @@ enum Enum {
 //~^^^^^ ERROR
 //~^^^^^^ ERROR
 //~^^^^^^^ ERROR
+//~^^^^^^^^ ERROR
    }
 }
 

src/test/compile-fail/deriving-span-PartialOrd-enum.rs

@@ -27,6 +27,7 @@ enum Enum {
 //~^^^^^ ERROR
 //~^^^^^^ ERROR
 //~^^^^^^^ ERROR
+//~^^^^^^^^ ERROR
      )
 }
 

src/test/compile-fail/deriving-span-PartialOrd-struct.rs

@@ -26,6 +26,7 @@ struct Struct {
 //~^^^^^ ERROR
 //~^^^^^^ ERROR
 //~^^^^^^^ ERROR
+//~^^^^^^^^ ERROR
 }
 
 fn main() {}

src/test/compile-fail/deriving-span-PartialOrd-tuple-struct.rs

@@ -26,6 +26,7 @@ struct Struct(
 //~^^^^^ ERROR
 //~^^^^^^ ERROR
 //~^^^^^^^ ERROR
+//~^^^^^^^^ ERROR
 );
 
 fn main() {}

src/test/compile-fail/issue-3344.rs

@@ -10,7 +10,7 @@
 
 #[deriving(PartialEq)]
 struct thing(uint);
-impl PartialOrd for thing { //~ ERROR not all trait methods implemented, missing: `lt`
+impl PartialOrd for thing { //~ ERROR not all trait methods implemented, missing: `partial_cmp`
     fn le(&self, other: &thing) -> bool { true }
     fn ge(&self, other: &thing) -> bool { true }
 }