Fix netcdf issue for Hamtide tests

[SorooshMani-NOAA]

Tests that required Hamtide failed due to an issue with the latest version of netcdf library installed using pip. Fixes #188

[SorooshMani-NOAA]

@WPringle I tried to address a couple of issues for tests to run successfully in this PR locally everything works, but now I see errors in macos version! When I look at the diff (I outputted the diff in the test for debugging) I see some section of tides that are different from the reference (on linux on my machine it's fine).

If you're working on a mac can you please run tests and see what you get?

[WPringle]

@SorooshMani-NOAA On Argonne linux machine using python 3.10 I'm getting a failed test for test_stations.py

[SorooshMani-NOAA]

I had to update the station references. For some reason the order of stations changed (maybe due to a change in searvey or even server side).

Make sure you have the latest reference files for local testing. What about macos? Were you able to test on a mac?

[SorooshMani-NOAA]

Thanks to @AtiehAlipour-NOAA we ran the tests on a mac machine and we get a crash from shapely:

The run env was created by conda with python=3.10, gdal netcdf4 and hdf5 installed using mamba/conda and the rest of packages installed using pip.

git clone https://github.com/oceanmodeling/adcircpy
cd adcircpy/
git checkout test/fix_hamtide_nc
mamba create -n adcircpytest -cconda-forge python=3.10 netcdf4 hdf5 gdal
conda activate adcircpytest
pip install -e .[testing]
pytest tests/
ls tests/data/output/

If we don't install gdal by conda we'll get an error:

On the CI for mac brew is used for netcdf and hdf, but the rest are installed by pip:

brew install hdf5
brew install netcdf

[SorooshMani-NOAA]

After more testing via updating tests and ref files it seems the reason behind the difference is how grid-data linear is results are different between mac and linux, but I'm not sure why! See code state at 82df7cd where all calculations for point location and amp/phase are done within the test code.