Wrong score from assess_reverse_dependencies

[albertopessia]

The documentation of assess_reverse_dependencies states:

The more packages that depend on a package the more chance for errors/bugs to be found

A package with no reverse dependencies should then get a maximum score of 1 while a package with many reverse dependencies should get a score close to zero.
At the moment assess_reverse_dependencies is doing the opposite and does not produce values in the range [0, 1] but rather on [0.2463761, 1].

First of all, the logistic function should be decreasing therefore the function implemented by metric_score.pkg_metric_reverse_dependencies should have an opposite sign growth rate.

Also, the logistic function is defined on the whole real line while the number of packages variable is non-negative. If number of packages is not log-transformed, then the asymptote at x -> -Inf will never be reached (see the minimum score of 0.2463761).

I suggest using the log-logistic function because uses the same scale of number of packages and it is well defined also when number of packages is zero:

metric_score.pkg_metric_reverse_dependencies <- function(x, ...) {
  1 - 1 / (1 + (5 / length(x))^2.5)
}

The growth rate of 2.5 is just an example and it should be tuned to the desired score behavior.