Should run_app() return the shiny object or print the output?

[ColinFay]

This is in the spirit of #146 and #136

The current run_app() function does not "actually run the app". It does create an app object, which is returned and launched by the print() method for Shiny objects.

This is problematic for example for profvis, which doesn't work "natively" on xxx::run_app()

One way to do this would be to change run_app to

run_app <- function(...) {
  app <- with_golem_options(
    app = shinyApp(ui = app_ui, server = app_server), 
    golem_opts = list(...)
  )
  print(app)
}

which would allow profiling.

Though, this wouldn't allow to

z <- xxx::run_app()

so the object is no longer available to be manipulated.

Which seems like a good tradeoff, considering that we are already manipulating the app inside the run_app() function.

[ColinFay]

with_golem_options now explicit calls print on the app object, solving some issues with benchmarking the application. This explicit print can be turned off by setting print to FALSE in with_golem_options

[ColinFay]

Closed via #496