Bug fix keyword encoding when there are multiple keywords with '&'

[ThiesDS]

See issue, as it fixes #300.

[PMassicotte]

Can you just use lapply?

[ThiesDS]

As the new line uses the outcome of the previous line (recursively), I don't think lapply is applicable. But if you want I can write it as for loop.

[PMassicotte]

This is weird, form the help file:

sub and gsub perform replacement of the first and all matches respectively.

[ThiesDS]

Right. That is why I wrote "It turns out that this only converts '&' to '%26' in the last keyword used - still don't know why" in the issue. I think this has to do with the regexp and the perl = TRUE. If I use the gsub with perl = FALSE it throws and error saying that "(?:\\G(?!^)|\\[\\s*)[^][\\s]*\\K\\&(?!])(?=[^][]*])" is not a regexp. However, I'm really not an expert in regexp.