Adapt write_dwc() for use by OpenCPU

[PietrH]

Like download_acoustic_dataset() see #32 write_dwc() doesn't exclusivly return an object, if directory is not NULL it'll write out a csv.

Unlike download_acoustic_dataset(), the handling is more simple for write_dwc(). A project object is created from get_animal_projects() which is used to create an object dwc_occurrence which is either returned or written out as csv.

dwc_occurrence is currently created via a locally stored sql query which is edited in place based on the project object. Later on we might switch over to darwincore mapping via dplyr similar to inbo/camtraptor#207, but as long as we stick to the sql mapping, it doesn't make much sense to split this function up into multiple requests.

Thus at least initially, I'll implement this as a single API call that returns dwc_occurrence, the client can then either return or write out based on the directory argument.

TODO

• adapt test-write_dwc.R: expect dwc_occurrence, not csv
• switch over to using credentials
• remove directory argument
• update documentation
• can we forward messages? progress reports?
• remove writing to csv
• pass tests
• pass R CMD CHECK
• fix broken examples #35

[PietrH]

Currently the function prints a message to console:

Reading data and transforming to Darwin Core.

Right after fetching the imis dataset id and title and starting the query to fetch the darwincore occurrence data. This behaviour could be maintained if we split off this first query to the client side.

This would however completely change the input arguments to write_dwc() between the version in etnservice and in etn, creating an exception for any future generic sql helper for etnitself. Or for the option of inheriting documentation betweenetnandetnservice`.

For now we'll keep it simple, and drop the message instead.

[PietrH]

general progress is tracked in inbo/etn#281