Try with hdf5 pin

[hoxbro]

This seems to have been the culprit for the no-solve.

The steps I took to find out this package created the problem.

1. create an environment:

mamba create -n tmp35 python=3.7 pyctdev -c pyviz/label/dev  -c conda-forge -c nodefaults
mamba activate tmp35
mamba install -y  "param" "pyct" "setuptools" -c pyviz/label/dev  -c conda-forge -c nodefaults

2. Trying to run the same code as the CI, wich gave the no solve.

mamba install   "colorcet" "dask-core" "datashape" "numba >=0.51" "numpy" "pandas" "param" "pillow" "pyct" "requests" "scipy" "toolz" "xarray" "codecov" "fastparquet" "flake8" "nbconvert" "nbformat" "nbsmoke >0.5" "netcdf4" "pyarrow <11" "pytest" "pytest-benchmark" "pytest-cov" "rasterio" "rioxarray" "spatialpandas" "bokeh <3.0" "geopandas" "holoviews" "matplotlib" "scikit-image" "spatialpandas"  -c pyviz/label/dev  -c conda-forge -c nodefaults --dry-run --offline

3. I removed all the geo-related packages, which could solve.

mamba install   "colorcet" "dask-core" "datashape" "numba >=0.51" "numpy" "pandas" "param" "pillow" "pyct" "requests" "scipy" "toolz" "xarray" "codecov" "fastparquet" "flake8" "nbconvert" "nbformat" "nbsmoke >0.5" "netcdf4" "pyarrow <11" "pytest" "pytest-benchmark" "pytest-cov" "bokeh <3.0" "holoviews" "matplotlib" "scikit-image"   -c pyviz/label/dev  -c conda-forge -c nodefaults

4. Installed the geo-packages, which could solve, but gave these downgrades:
   

5. I Tried to pin the downgrades one at a time and saw that hdf5 was a problem.

[ianthomas23]

Thanks @hoxbro, this is great!