[FEATURE] - 'Open' MF6sim after running to facilitate access to functions like open_head and the like

[jdelsman]

For large (mf6) models it is inconvenient to run them using the .run() function of MF6 (too risky within a script, preference to run on a different computer than generating input etc). That means that the .open_head etc functions cannot be used, and the low-level functions must be used.

Describe the solution you'd like
I'd like to have a light way of opening a mf6sim object after (different script) the model has been run, so the open_head etc functions are easily available. Like a .from_file, with an additional 'rundir' argument or something. But preferably without much reading input data overhead, just what is necessary (dis?). And allowing for parallel calculations.

[Huite]

Making sure the problem description is clear: are the low level methods really that "low level"?
Or is the problem mostly the parallel output?

I'm asking because the open methods in principle currently require two arguments: paths to the head file and the GRB file, and this can (only) be reduced to a single argument. Admittedly, functions with one argument are better than functions with two arguments, and convenient support for parallel output means a single directory argument anyway...

Using DIS instead of GRB is best seen as a separate issue, in my view.

Are there other uses for a lightweight MF6 sim object? It is just output?

Note that if the input is in netCDF form, then "opening" the data via xarray is lightweight by default. I don't think we've investigated the current netCDF capabilities much.