Suggestion to retain chr and fct vectors in merge_samples

[roey-angel]

Posted as a response to issue #893 but seem to have gone unnoticed.
What about applying a mean to all numeric variables in sample_data() and something like summarise_all(list(~paste(unique(.), collapse = ","))) to all other?
This will retain all unique chr and fct values and in case of conflicting entries for merged samples, both will be retained and separated by a ",".

In the meantime, I use this function to replace merge_samples()