This provides a slightly better way to write functionality that operates
on boot environments, in that it does some validation prior to executing
the subcommand and sets some variables up in the environment to describe
the currently-selected BE root.
My initial idea for a candidate was something like bectl ingest,
which can then be written easily like (with arg validation added):
mntp=$(bectl -r "$BE_ROOT" mount "$1") tar -C "$mntp" -xvf - bectl -r "$BE_ROOT" umount "$1"
This lets bectl cope with the BE root validation such that we only need
to glue a few things together and avoid making assumptions about the
pool, and we can provide a semi-consistent interface for things intended
to operate on boot environments.