The Vector implementation we're currently writing can grow by one element in
amortized constant time (meaning that any n appends take O(n) time), but each
individual insertion is not guaranteed to run in O(1) time.  This is because if
we run out of space and need to reallocate the internal array, we need to do
O(n) work to copy the elements over.

Come up with a different way of implementing the Vector so that at, push_back,
pop_back, size, and empty all run in worst-case O(1) time.  This doesn't mean 
that they all have to take the same amount of time, but rather that the runtimes
don't depend on the size of the Vector.  You only need to implement the following
functions, but you can do more if you'd like:

push_back  pop_back  at  size  empty