update(Mf6Splitter): change how node mapping is stored and loaded

[jlarsen-usgs]

• save_node_mapping now writes a compressed hdf5 file and stores additional modelgrid reconstruction information for original and split model representations
• load_node_mapping changed to a static method that returns a Mf6Splitter object that can be used for array reconstruction
• updated autotests
• added example usage to mf6_parallel_model_splitting_example.py
• json node mapping files no longer supported, user must rewrite node mapping files as hdf5 files.
• bugfix for observation file naming convention on remapped obs files

[jlarsen-usgs]

@mjr-deltares, here are the updates to the model splitter for more efficient storage and loading of the "node mapping" files. If you'd like to give them a test, usage is as follows

mfs = Mf6Splitter(my_sim)
new_sim = mfs.split_model(array)

node_map_file = "my_node_map.h5"
mfs.save_node_mapping(node_map_file)

mfs2 = Mf6Splitter.load_node_mapping(node_map_file)

I tested this on the average annual long island model and HDF5 file processing and write time was ~5 seconds, read and reconstruction time was ~1.5 seconds, and HDF5 file size was about 63 MB, which includes information to reconstruct the original and split model grids, and the mappings to link these together.

[codecov]

Codecov Report

Attention: Patch coverage is 96.15385% with 8 lines in your changes missing coverage. Please review.

Project coverage is 74.8%. Comparing base (bb9824e) to head (e6afa43).
Report is 68 commits behind head on develop.

Files with missing lines	Patch %	Lines
flopy/mf6/utils/model_splitter.py	96.5%	7 Missing ⚠️
flopy/plot/map.py	50.0%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff            @@
##           develop   #2465     +/-   ##
=========================================
+ Coverage     68.4%   74.8%   +6.4%     
=========================================
  Files          294     296      +2     
  Lines        59390   62736   +3346     
=========================================
+ Hits         40652   46977   +6325     
+ Misses       18738   15759   -2979     

Files with missing lines	Coverage Δ
flopy/mf6/utils/output_util.py	74.4% <100.0%> (+2.0%)	⬆️
flopy/plot/map.py	82.1% <50.0%> (+2.0%)	⬆️
flopy/mf6/utils/model_splitter.py	74.0% <96.5%> (+15.2%)	⬆️

... and 254 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[mjr-deltares]

@mjr-deltares, here are the updates to the model splitter for more efficient storage and loading of the "node mapping" files. If you'd like to give them a test, usage is as follows

mfs = Mf6Splitter(my_sim)
new_sim = mfs.split_model(array)

node_map_file = "my_node_map.h5"
mfs.save_node_mapping(node_map_file)

mfs2 = Mf6Splitter.load_node_mapping(node_map_file)

I tested this on the average annual long island model and HDF5 file processing and write time was ~5 seconds, read and reconstruction time was ~1.5 seconds, and HDF5 file size was about 63 MB, which includes information to reconstruct the original and split model grids, and the mappings to link these together.

Hi @jlarsen-usgs , I forgot to answer you last week but it's on my radar now. Thanks for making these changes!

[jlarsen-usgs]

ssm updates included in PR. Closes #2474