===================
* Change the `--scale=r` parameter in `sample para` and `sample mono` to
represent `n^(1/r)` instead of `n^r`. With this parameterization, `r` ranges
from 1 to infinity and thus does not conflict with the `--proportion` term
(which ranges from 0 to 1). They do overlap at `--scale=1` and
`--proportion=1`, but these are equivalent, both sampling everything.