add slev retrieval from ddl

[veenstrajelmer]

TODO:

• DDL API interaction functions via in ddlpy
• add ddl to subset/retrieve insitu slev functions
• notebook: add ddl
• notebook: catch empty gdf at .iloc[0] line with if subset_gpd.empty: continue)
• notebook: prevent ddl prints
• remove "old" hatyan code and hatyan dependency
• Add support for time strings in ddlpy.measurements() ddlpy#39
• add function to simplify measurements dataframe ddlpy#42
• Improvements for measurements dataframe ddlpy#38
• less ugly station_name_unique method (duplicate station codes, what is what? >> we use station_name_unique to avoid overwriting) >> disabled
• _make_hydrotools_consistent(ds) for all sources, run testbank to see if required attrs are present.
• add more todo from P:\11210366-003-getijverandering\insitu_data\RWS_DDL\download_ddl.py and dfm_tools/observations.py
• rename ddl to rws or rwsddl (no dash)
• remove meta_dict from retrieve function and add exception to read_catalog function if anything was passed that could not be filtered. This cleans up code significantly.
• optional (can also be done later): assert for presence of nc-file in retrieve test, will currently fail for some sources since requested time/station is not available. test_ssh_retrieve_data does not always download a file, assert for this and try different station if no file (maybe different time min/max)
• rerun example notebook
• move station loop and file saving (incl _make_hydrotools_consistent) to generic part to avoid duplication
• add crs wgs84 to all catalog dataframes (or is that the default?), test this by applying .to_crs() to dataframe
• waterlevels as float32 instead of float64
• hydrotools-compatible function: add assert for time variable and dimension (both lowercase)
• maybe clean cmems dataset if hydrotools cannot work with lat/lon/pos dim
• download cmems raw file to tmpdir so it does not have to be deleted
• fix failing tests/test_observations.py::test_ssh_retrieve_data[rwsddl] testcase on github, probably same cause as Can only query a limited date range from the API ddlpy#18, so can probably be fixed by sorting locations dataframe on numericid column in test (and possibly look for a new index, or get single row via station-code instead to be more future proof)
• move adding station_name/lat/country/etc attrs to general function, requires these attrs to be present in the location catalog/df already, which is desireable. Assert for this.
• assert for empty dataframe instead of none
• notebook: add plot_borders()
• add test for rwsddl_ssh_get_time_max()
• Create ddlpy 0.3.0 release ddlpy#56
• update ddlpy dependency from git link to pypi package (>=0.3.0), make non-optional and remove import checks from observations.py (move import to top of script)
• rerun notebook