feat(pathlike): support pathlike throughout public interface

[wpbonelli]

Working toward #1729, continuation of #1712, #1603, and #1583

• update code where needed to support os.Pathlike in user-facing API
• update docstrings and type hints
• update most tests to use Path instead of str
• add tests checking both Path and str for core classes/utilities
• addverbose flag to more utility methods to optionally enable/suppress output
• update some notebooks to use pathlib, but not all
• add sim_path property to MFSimulation as a shortcut to simulation_data.mfpath.get_sim_path()
  • guaranteed Path whether user provides workspace as str or PathLike
  • update tests and mf6 tutorial notebook to exercise/demonstrate
• preserve str properties where this was already the convention
  • e.g. BaseModel.model_ws
  • could consider refactoring to Path in future, but may break usages
• update resolve_exe function, add tests, and use it from both BaseModel as well as run_model()

[codecov]

Codecov Report

Merging #1730 (9ac6f66) into develop (cf3a517) will increase coverage by 0.1%.
The diff coverage is 82.0%.

@@            Coverage Diff            @@
##           develop   #1730     +/-   ##
=========================================
+ Coverage     71.9%   72.1%   +0.1%     
=========================================
  Files          253     253             
  Lines        56004   56043     +39     
=========================================
+ Hits         40282   40416    +134     
+ Misses       15722   15627     -95     

Impacted Files	Coverage Δ
flopy/plot/map.py	82.3% <ø> (ø)
flopy/utils/formattedfile.py	88.8% <ø> (-0.2%)	⬇️
flopy/utils/compare.py	40.7% <42.8%> (+10.3%)	⬆️
flopy/utils/zonbud.py	66.7% <66.6%> (-0.1%)	⬇️
flopy/export/utils.py	65.3% <72.3%> (+0.1%)	⬆️
flopy/mbase.py	69.7% <81.9%> (+0.4%)	⬆️
flopy/utils/datafile.py	74.0% <83.3%> (+0.1%)	⬆️
flopy/utils/modpathfile.py	86.0% <83.3%> (+<0.1%)	⬆️
flopy/utils/gridgen.py	84.4% <88.8%> (+0.1%)	⬆️
flopy/export/shapefile_utils.py	90.0% <90.0%> (-0.3%)	⬇️
... and 29 more

[mwtoews]

Thanks, looks good so far, see following comments and suggestions. Are there a few representative tests that use str remaining?

[wpbonelli]

Added/expanded a few tests covering both str and Path for Modflow and MFSimulation read/write, gridgen, some export utilities. Planning to add a few more.

[mwtoews]

Nice stuff, getting closer, see attached suggestions/comments. I'm primarily looking at the main module, and have not seen many of the edits to notebooks or autotests.

[christianlangevin]

Looks great @w-bonelli. Nice addition. As you mentioned, a few remaining os.path.join remain in notebooks, but we can get those later.

[christianlangevin]

Out of curiosity, why is str needed here?

[wpbonelli]

it's not strictly — previously output_filename was a str, switching to Path would require changes to L160 and L523 and I figured it may be safer to keep it a string in case any external usages had similar expectations

[wpbonelli]

@mwtoews thanks for the catches, I think they've all been fixed.

[mwtoews]

Just a few (probably) final comments. And again, I'm only reviewing the changes in flopy/*

[mwtoews]

Currently this is a required parameter, because it does not have a default e.g. = None.

[wpbonelli]

The docstring indicated namefile=None could be used for models that don't require namefiles — the function seems to handle this case OK, just omits it from the args passed to the executable

[wpbonelli]

Fixed more docstrings/typehints, thanks @mwtoews, also updated exe resolve function introduced in #1727

• add docstrings
• test cases described in bug: location of mf6 executable using relative path #1633
• log externally (factor silent and model_ws params out)
• name resolve_exe_name -> resolve_exe (full path is returned if successful)

[mwtoews]

Thanks, looks good! I only have a minor suggestion remaining. This PR could be merged soon.