I thought about this some more, and I believe the Writer-based approach
in the previous version of the LargeObject API does not scale: It makes
it very hard to write code that uploads segments without resorting to a
buffer the same size as the segments. I don't want gigabyte-scale
buffers filling up my RAM, so this commit switches to a different API
based on Readers. LargeObject.Append() now behaves very similar to
Object.Upload(), which I find quite nice.

1. Move common methods of AccountHeaders, ContainerHeaders,
   ObjectHeaders into a base type Headers.

2. Fold Headers into RequestOptions to remove one of the two optional
   arguments on request methods. The new Headers.ToOpts() method
   offers a nice experience for users passing Headers to request
   methods.

   The Update() methods keep the explicit Headers argument since the
   Headers argument is not optional there.

   The only downside is that we lose a bit of type-safety because
   RequestOptions takes any Headers instance, so e.g. ContainerHeaders
   could be passed to Object.Upload(). I believe the benefits outweigh
   this problem.

* Gophercloud dependencies move into subpackage gopherschwift.
* Tests move into subpackage tests (to avoid import cycles).
+ Rename "Client" to "Backend".