Average data with more than just id

[estebanbeck]

When you try to do an average with more than one "by" you brake the behavr object and you end up with no key.
This is the issue and a potential patch to re-attach the metadata to the resulting summary.

metadata <- data.frame(id = paste0("toy_experiment|",1:2),
                       condition = c("A", "B"))
dt_test <- toy_activity_data(metadata = metadata)
#to add day number, and light phase
dt_test [,day:=floor(t/days(1))]
dt_test [,phase:=ifelse(t %% hours(24)>hours(12),"Dark","Light")]
dt_test [,phase:=factor(phase, levels= c("Light","Dark"))]
dt_summary_test  = dt[,
                  .(var = mean(asleep))
                  ,keyby=c("id", "phase")]
dt[meta=T][dt_summary_test]