Handling data when building flopy packages from model dataset

[dbrakenhoff]

This PR aims to inform users how nlmod is dealing with data when users build flopy packages.

In terms of variables we're trying to stick to flopy default names as much as possible. This is mostly the case, though some differences might remain. To distinguish datasets between different packages the convention we're using is <package-abbreviation>_<variable-name>.

Basically, there are two options:

1. variables are passed as strings, and point to an array in the dataset
2. variables are passed as arrays (data arrays, or classic numpy arrays, or constants).

Option 2 will have precedence over option 1, so passing your own data-array will always override any stored dataset (with the same name).

Option 1:

• notify user data is read from dataset (debug level)
• notify user if dataset does not exist and return None (warning)

Option 2:

• if a data array is passed, notify user if data with the same name is stored in dataset (info)
• if any other datatype, just use passed value (e.g. a numpy array), (no message)

The function _get_value_from_ds_datavar takes care of this behavior for data variables. The function _get_value_from_ds_attr does something similar for attributes. Both these functions are tested in test_003_mfpackages.py, which can also be helpful in understanding how to call these methods.

Other things worth mentioning:

• deprecated da_name keyword arguments for nlmod.gwf.ghb and nlmod.gwf.drn packages
• deprecated chd keyword argument, renamed to mask in nlmod.gwf.chd
• rename "_peil" data variable to "_stage" for RWS surface water data

[dbrakenhoff]

CI error is unrelated to this PR (something about data from Wetterskip Fryslan), so this is ready for review.

[rubencalje]

Looks good. There are probably some other places we need to make input more consistent (I cannot name one though), but let's merge these improvements!