Benchmark different operations in ‘box’

[klmr]

As discussed in #262, there should be a suite of benchmarks of different operations, and how they compare to equivalent base R operations. The results should be published as a vignette.

Measure performance of …

1. box::use(pkg)
   1. When ‘pkg’ is already loaded
   2. When ‘pkg’ isn’t loaded
   3. When ‘pkg’ isn’t loaded, but its dependencies are
2. box::use(pkg[...])
   1. When ‘pkg’ is already loaded
   2. When ‘pkg’ isn’t loaded
   3. When ‘pkg’ isn’t loaded, but its dependencies are
3. library(pkg)
   1. When ‘pkg’ is already loaded
   2. When ‘pkg’ isn’t loaded
   3. When ‘pkg’ isn’t loaded, but its dependencies are
4. box::use(./mod)
   1. When ./mod is already loaded
   2. When ./mod isn’t loaded
   3. When ./mod isn’t loaded, but its dependencies are
5. box::use(path/to/mod)
   1. When ./mod is already loaded
   2. When ./mod isn’t loaded
   3. When ./mod isn’t loaded, but its dependencies are
6. box::use(./mod[...])
   1. When ./mod is already loaded
   2. When ./mod isn’t loaded
   3. When ./mod isn’t loaded, but its dependencies are

And …

1. mod$name
2. name, for an attached name
   1. Locally, in the same environment
   2. Locally, in the parent environment
   3. In the global search path