magrittr::set_names and rlang/purrr::set_names are inconsistent

[mmuurr]

magrittr::set_names conflicts with rlang/purrr::set_names in an incompatible way.
(The latter version requires only a single argument, using the first arg as the default for the second arg.)

Given the tight integration of magrittr in the tidyverse, should these inconsistencies be snuffed-out?

[smbache]

Yes. Magrittr aliases should be removed - and will be at some point.

[mmuurr]

Yay, thanks!

(In the meantime, my work-around is what I've dubbed the 'owl idiom', as it returns the same results in both cases: %>% set_names(.,.).)

[hadley]

Part of #108