Concatenate salary files in `tidy/`

[Thibauth]

Is there any reason to keep the salary files separated in tidy/? Are they semantically different? For other datasets, we concatenate all the records as part of the cleaning process. I believe this falls under the general philosophy of what we do in the cleaning step: it is non-controversial, and anyone who wants to work with the raw data would want to concatenate these files anyway.

[trevorcampbell]

I think this should be fine. I had originally kept them separate because they come from three separate xls files (each containing multiple sheets, one for each year). I believe they have the same fields so this shouldn't be hard.

In addition, we'll need to change Makefile and merge_salary.py to reflect that change. Will leave this as an open TODO for now.