bug: gwf.modeltime.steady_state inconsistent with input/output

[aleaf]

Describe the bug
The gwf.modeltime.steady_state attribute appears to be returning True for every period entered in the steady_state argument to ModflowGwfsto, regardless of whether the entry for that period is True or False. The actual STO input file written by Flopy does not seem to be affected by this.

To Reproduce

import flopy as fp
out_dir = 'tmp'
sim = fp.mf6.MFSimulation(sim_name='testsim',
                   version='mf6',
                   sim_ws=out_dir)
tdis_rc = [(1.0, 1, 1.0), (10.0, 120, 1.0), (10.0, 120, 1.0), (10.0, 120, 1.0)]
tdis = fp.mf6.ModflowTdis(sim, nper=len(tdis_rc),
                          perioddata=tdis_rc)
ims = fp.mf6.ModflowIms(sim)
gwf = fp.mf6.ModflowGwf(sim, modelname='testmodel')
dis = fp.mf6.ModflowGwfdis(gwf)
sto = fp.mf6.ModflowGwfsto(
    gwf,
    pname="sto",
    save_flows=True,
    iconvert=1,
    ss=1e-4,
    sy=0.2,
    steady_state={0: True, 1: False, 2: False, 3: False},
    transient={0:False, 1: True, 2: True, 3: True},
)
gwf.modeltime.steady_state

array([ True,  True,  True,  True])

Expected behavior

array([ True,  False,  False,  False])

More broadly, since a stress period can't be both steady-state and transient, maybe it would be better to refactor the Flopy input to a single steady_state dictionary of True/False values. Is there any reason why we wouldn't want to do this?

[aleaf]

A related issue with loading a model: If the STO package isn't included in the load, gwf.modeltime.steady_state returns all True for each stress period.

sim.write_simulation()
loaded_sim = fp.mf6.MFSimulation.load('testsim', sim_ws=out_dir, load_only=['dis'])
loaded_gwf = loaded_sim.get_model('testmodel')
loaded_gwf.modeltime.steady_state

array([ True, True, True, True])

Maybe loaded_gwf.modeltime.steady_state should be None at this point to make it clear that the model doesn't have this information?

[spaulins-usgs]

The first problem should now be fixed.

Regarding your second comment, the MF6 documentation states, "If the STO Package is not included for a model, then storage changes will not be calculated, and thus, the model will be steady state.". Given this, I think it is reasonable for flopy's modeltime.steady_state to be true when the STO package does not exist.