Improve docstring

Author: adrhill

src/input_augmentation.jl

@@ -91,7 +91,7 @@ A wrapper around analyzers that augments the input with `n` samples of additive
 This input augmentation is then averaged to return an `Explanation`.
 
 Defaults to the normal distribution `Normal(0, std^2)` with `std=1`.
-For optimal results, $REF_SMILKOV_SMOOTHGRAD recommends setting `std` between 10% and 20% of the input range of every sample,
+For optimal results, $REF_SMILKOV_SMOOTHGRAD recommends setting `std` between 10% and 20% of the input range of each sample,
 e.g. `std = 0.1 * (maximum(input) - minimum(input))`.
 
 ## Keyword arguments