construct pyIAST `ModelIsotherm` without data, using only isotherm parameters

parametric_ModelIsotherm_constructor.py

## for example, construct a Lanmguir model with params M, K
def construct_langmuir_model(M, K):
    df_bogus = pd.DataFrame([[1.0, 1.0]], columns=["P_bogus", "N_bogus"]) # bogus data
    m = pyiast.ModelIsotherm(df, model="Langmuir", pressure_key="P", loading_key="N") # will fit Langmuir model with bogus params
    # overwrite bogus model params
    m.params['M'] = M
    m.params['K'] = K
    # overwrite bogus data to be sure it is never used.
    m.df = pd.DataFrame([[np.NaN, np.NaN]], columns=["P_bogus", "N_bogus"]) # bogus data
    return m
  
  # some tests to convince yourself it's all right
  m = construct_langmuir_model(1.0, 10.0) # M = 1, K = 10
  m.loading(1000.0) # n(inf) = M
  m.loading(1 / 10.0) # n(1/K) = 0.5 M
  m.df # should have NaN's to clarify this is bogus data