Unformat package names at Acquisition function vignette (#931)

Use tidy style in Acquisition function vignette. Fixes #922

vignettes/acquisition_functions.Rmd

@@ -91,7 +91,7 @@ knitr::include_graphics("figures/trade_off_20.svg", auto_pdf = FALSE)
 
 There are two main strategies for _dynamic trade-offs_ during the optimization: 
 
- * Use a function to specify the parameter(s) for the acquisition functions. For expected improvement, this can be done using `exp_improve(trade_off = foo())`. `foo()` should be a function whose first parameter is the current iteration number. When `tune` invokes this function, only the first argument is used. A good strategy might be to set `trade_off` to some non-zero value at the start of the search and incrementally approach zero after a reasonable period. 
+ * Use a function to specify the parameter(s) for the acquisition functions. For expected improvement, this can be done using `exp_improve(trade_off = foo())`. `foo()` should be a function whose first parameter is the current iteration number. When tune invokes this function, only the first argument is used. A good strategy might be to set `trade_off` to some non-zero value at the start of the search and incrementally approach zero after a reasonable period. 
 
  * `control_bayes()` has an option for doing an additional _uncertainty sample_ when no improvements have been found. This is a technique from the active learning literature where new data points are sampled that most help the model. In this case, the candidate points are scored only on variance and a candidate is chosen from a set of the _most_ variable design points. This may find a location in the parameter space to help the optimization make improvements.