This PR is fairly hefty.

A Segment class is implemented to serve as the basic building block for the Pipeline. This clarifies the link between the fundamental make_*() functions and the Pipeline object. In particular, each make_*() call constructs a Segment which is appended to the Pipeline.

The Segment class has an execute() method, which replicates the execution behaviour of the make_*() functions. Hence Segment$execute() can be run to rebuild the targets associated with a given segment without adding a new segment to the Pipeline as would be done if we called make_*() again.

This makes it straightforward to implement a Pipeline$build() method: we simply sort the Segments topologically and then call execute() on each in turn.

The Segment contains all the information that the old makepipe_result S3 class did and therefore makes the latter redundant. The execution metadata (e.g. execution result, execution time, etc.) is updated via the Segment$update_result() method whenever Segment$execute() is called