New format drivers

[jaideep777]

This PR implements a flexible downsampling routine that takes 4 options: legacy, 24hr, daytime, 3hrmax. It facilitates writing all of these in the rsofun drivers dataset. It is currently fully backwards compatible.

The revised downsampler and rsofun formatter are implemented in *_phydro.R files. This is because the original fdk_downsample_fluxnet and fdk_format_drivers scripts are used also elsewhere in the code.

I suggest you cleanly separate the downsampling step into a numbers analysis step after the fdk data collection step. Currently, the data collection step writes both hh and dd files. It might be better if it wrote only hh files, and dd files are generated in the separate downsampling step.

[stineb]

Thanks! It would be safer if your point about the separated downsampling is implemented in your fork before merging the pull request. We will have to avoid the files with _phydro. Additional functionalities should be integrated into the existing functions, possibly making them optional, while optional backward compatibility ideally enabled.