Find a more elegant solution to selecting variables that do not exist

[WillemSleegers]

In some of our functions we select variables (e.g., stats_to_df()), but the variables do not always exist. Currently we solve it by using one_of() in the select(), but it throws a warning when a variable isn't found.

We should see if we can find a solution that doesn't throw a warning.

[WillemSleegers]

This has become less pressing with the new tidier implementation. However, it might still occur (for example when there are only t-tests).

[WillemSleegers]

Fixed this.