pathing issues with models in sub directories

[briochh]

This is possibly related to #1115.

I understand that external files within packages are expected (by modflow6) to be pathed relative to the simulation directory. I think flopy is trying to handle this where models are in subdirectories (or potentially just different directories to the simulation directory). The strategy seems to be that flopy prepends the model relaitve path to the file entry within the package.

For the most part this seems to work . However, there are a couple of places where things seem to get confused:

1. Observation Files are not prepended with the relative path -- this isn't to much of a biggy as if using the obs.initialize() method one can pass the full rel path explicitly to the filename arg.
2. External files are (re)prepended with the model relative path when loading the simulation back into flopy -- for e.g. the path to delc which in the model file is model_d/delc.txt becomes model_d/model_d/delc.txt in the m.dis.delc file entry after loading the sim and model back into flopy. Not sure if there is a work around that doesn't require package by package file by file stripping of the additional rel path.

urgh paths!

[briochh]

I think also related to #1126

[briochh]

I have plumbed in a simple test on my fork to catch this issue:
https://github.com/briochh/flopy/blob/437263b510c4a6f82a4fb0f41f76ef8273ae231d/autotest/t501_test.py
The issue seems to relate to the fact that we loop through the same methods when reading and writing.

When writing, we append the model relative path to the external file path (which is passed to flopy initially as relative to the model_ws). The same method must be being applied when reading but the true stored filename already include the relative path from the sim, so we end up with an extra model_rel_path. I would be happy to help some more but I can't work out where this appending is happening (I can't even work out where the entry is initally read!).