slot_map: Do not require Container to provide reverse_iterator typedefs.

[Quuxplusone]

Attn: @Masstronaut

With this patch, slot_map<T,K,C> will propagate the "reverse-iterability" of C<T>, both the existence of the member typedefs and the existence of the member functions rbegin, rend, crbegin, and crend (as a group, not individually).

Is this desirable? I tend to think yes, because implementing reverse-iterability for a custom container is usually wasted busywork. OTOH, it could alternatively be addressed by some library (either SG14 or Boost or otherwise) providing a reversible mix-in:

template<class Ctr>
struct reversible : public Ctr {
    using reverse_iterator = std::reverse_iterator<typename Ctr::iterator>;
    using const_reverse_iterator = std::reverse_iterator<typename Ctr::const_iterator>;
    reverse_iterator rbegin() { return reverse_iterator(this->end()); }
    reverse_iterator rend() { return reverse_iterator(this->begin()); }
    // ... and so on ...
};

template<class T> using reVector = reversible<TestContainer::Vector<T>>;
slot_map<int, std::pair<int,int>, reVector> sm;  // OK

BTW, the fact that this reVector alias template can be passed to a template template parameter is very cool. I was worried that it wasn't going to be that easy.

[mattreecebentley]

Seems reasonable to me - does anyone have any objections to this? I'm not familiar enough with the code to comment as to side-effects.