Use "dbl" for Doubles

Since both doubles and integers are numeric. It is confusing to label a data.table column of doubles as "num". In tibble, such a column is labelled as "dbl", which is more accurate. 

```
> is.numeric(1)
[1] TRUE
> is.numeric(1L)
[1] TRUE
> 
```