cmRemoveDuplicates: Store unique iterators instead of values.

There is no need to copy all of the values in the container in
order to determine uniqueness.  Iterators can be stored instead
and can be used with standard algorithms with custom comparison
methods.

This also means that we use less space in case the value_type size
is greater than sizeof(iterator).  That is common for std::string
which may require up to 32 bytes (libstdc++ 5.0 and MSVC at least).
With libstdc++ 4.9 and older, std::string is 8 bytes, so we likely
don't gain anything here.

Inspired-by: Daniel Pfeifer <[email protected]>

Source/cmAlgorithms.h

@@ -176,6 +176,12 @@ private:
   Range const& m_range;
 };
 
+struct IterLess
+{
+  template<typename It>
+  bool operator()(It const& a, It const& b) const { return *a < *b; }
+};
+
 }
 
 template<typename Iter1, typename Iter2>
@@ -264,8 +270,8 @@ typename Range::const_iterator cmRemoveMatching(Range &r, MatchRange const& m)
 template<typename Range>
 typename Range::const_iterator cmRemoveDuplicates(Range& r)
 {
-  typedef std::vector<typename Range::value_type> UniqueVector;
-  UniqueVector unique;
+  typedef typename Range::const_iterator T;
+  std::vector<T> unique;
   unique.reserve(r.size());
   std::vector<size_t> indices;
   size_t count = 0;
@@ -273,11 +279,12 @@ typename Range::const_iterator cmRemoveDuplicates(Range& r)
   for(typename Range::const_iterator it = r.begin();
       it != end; ++it, ++count)
     {
-    const typename UniqueVector::iterator low =
-        std::lower_bound(unique.begin(), unique.end(), *it);
-    if (low == unique.end() || *low != *it)
+    const typename std::vector<T>::iterator low =
+        std::lower_bound(unique.begin(), unique.end(), it,
+                         ContainerAlgorithms::IterLess());
+    if (low == unique.end() || **low != *it)
       {
-      unique.insert(low, *it);
+      unique.insert(low, it);
       }
     else
       {