Test Pak with repo2docker for faster R builds

There are a couple of lessons we've learned that could be applied to make repo2docker R builds faster.

• Use pak
• Install binaries whenever possible

This will change how install.R scripts for repo2docker are written, and you want to pass a complete list of all the packages to pak so it can solve the fastest way to install and do things in parallel. pak will either need to be installed in the base repo2docker for R, or every install.R will need

install.packages("pak", repos = sprintf(
  "https://r-lib.github.io/p/pak/stable/%s/%s/%s",
  .Platform$pkgType,
  R.Version()$os,
  R.Version()$arch
))

Support for 3rd party repositories is supported

pak::[repo_add](https://pak.r-lib.org/reference/repo_add.html)(rhub = 'https://r-hub.r-universe.dev') 

which is important for things like lasR, or pulling binaries from posit's repo.

install.packages(
  "lasR",
  repos = c("https://r-lidar.r-universe.dev", "https://cloud.r-project.org")
)

We should be able to test all this on our own with existing repo2docker, before upstreaming changes.
Unclear if Science Support or JupyterHub team should lead. cc: @aprilmj

Start by reviewing the existing install.R scripts and the base repo2docker R setup, then test pak and binary installation with representative package lists. Compare build behavior before and after the experiment, including third-party repositories such as lasR. Done means the approach is validated locally and the required install.R or base-image changes are clear enough to upstream.