refactor(createpackages): use jinja for mf6 module code generation

[wpbonelli]

Carved out of #2317. Just move to Jinja as a first step. Type hints can come afterwards as they will take some effort to get right. The aim here is to match the old createpackages.py capabilities without relying on mfstructure.py. This is an initial step to trim mfstructure.py in favor of a leaner representation of the input spec. mfstructure.py is still used at runtime which will need to be unraveled in followup work.

Introduce a flopy.mf6.utils.codegen module with some new machinery for generating code with Jinja. This includes a bunch of filters handling quirks of the generated classes which we can aim to gradually eliminate. The templates should also get simpler over time.

The codegen module consumes TOML rather than legacy DFNs. Devtools is used at code generation time to first convert DFNs to TOML. This happens under the hood, with no change in apparent behavior to the user of the codegen utility.

Jinja2 and modflow-devtools are added to a new optional dependency group codegen. These are required to use the code generation utilities. These as well as transitive dependencies are added to environment.yml.

Miscellaneous:

• expand and add some mermaid diagrams to the MF6 module dev guide
• minor fix in flopy/mf6/data/mfdatastorage.py to avoid referencing a variable before it exists
• update DFNs as per refactor(dfns): add in_record attr to utl-ts and utl-tas vars MODFLOW-ORG/modflow6#2031

[deltamarnix]

I haven't looked through everything, but I left you a PR on your own branch with indications on how I would probably approach this. Let's take a look at it together and see if we can get rid of the shim :)

[codecov]

Codecov Report

Attention: Patch coverage is 4.91803% with 406 lines in your changes missing coverage. Please review.

Project coverage is 76.0%. Comparing base (bb9824e) to head (5c2a043).
Report is 48 commits behind head on develop.

Files with missing lines	Patch %	Lines
flopy/mf6/utils/codegen/filters.py	0.0%	308 Missing ⚠️
flopy/mf6/utils/codegen/__init__.py	0.0%	70 Missing ⚠️
flopy/mf6/utils/codegen/component.py	0.0%	20 Missing ⚠️
flopy/mf6/utils/generate_classes.py	40.0%	6 Missing ⚠️
flopy/mf6/utils/createpackages.py	50.0%	2 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff            @@
##           develop   #2333     +/-   ##
=========================================
+ Coverage     68.4%   76.0%   +7.6%     
=========================================
  Files          294     296      +2     
  Lines        59390   63717   +4327     
=========================================
+ Hits         40652   48474   +7822     
+ Misses       18738   15243   -3495     

Files with missing lines	Coverage Δ
flopy/mf6/data/mfdatastorage.py	75.4% <ø> (+6.5%)	⬆️
flopy/mf6/data/mfdatautil.py	77.9% <ø> (+23.3%)	⬆️
flopy/mf6/data/mfstructure.py	73.6% <100.0%> (+9.5%)	⬆️
flopy/mf6/mfbase.py	87.6% <100.0%> (+12.1%)	⬆️
flopy/mf6/mfpackage.py	81.9% <ø> (+17.3%)	⬆️
flopy/mf6/mfsimbase.py	75.6% <100.0%> (+13.4%)	⬆️
flopy/mf6/utils/createpackages.py	60.0% <50.0%> (+53.5%)	⬆️
flopy/mf6/utils/generate_classes.py	19.2% <40.0%> (+1.4%)	⬆️
flopy/mf6/utils/codegen/component.py	0.0% <0.0%> (ø)
flopy/mf6/utils/codegen/__init__.py	0.0% <0.0%> (ø)
... and 1 more

... and 246 files with indirect coverage changes

[deltamarnix]

I've only looked through some of the output files, but I have found some remarkable differences that we should check.

[wpbonelli]

Thanks. Yeah, this is why I want to wait to merge til after release. I think it'll take some time to ferret out all the minute differences and make sure no behavior has changed, even after merge.