invgamma 1.2

New features

invgamma has an updated README containing a simulation check on rinvgamma().

rinvgamma() now warns for shape less than 0.01, as it is unreliable there.
rinvchisq() now warns for df <= .01 and ncp <= 10, as it is unreliable there.
PDF functions now return 0 at x = 0 and -Inf at x = 0 on a log scale, consistent with the gamma distribution functions.
invgamma now cautions users about the parameterizations used in (d/p/q/r)invgamma() functions.
invgamma now has a MIT license.
Quantile functions now perform properly on a log scale, thanks to Keefe Murphy!

(d/p/q/r)invgamma() now defaults the rate parameter to 1, like (d/p/q/r)gamma().

(d/p/q/r)invchisq() and (d/p/q/r)invexp() are now included, tied to their base R counterparts.

dinvgamma() is now tied to dgamma() in base, instead of being a custom implementation of the density.