Reassess Gijmin

Currently, Gijmin is defined as 0.5 + 0.5 * np.tanh(1.25 * (-lower_limit + shares_dt) / 0.1).

This means that even when there is no lower limit whatsover, Gijmin has a large effect on small technologies. For instance, for lower_limit = 0.0 and shares_dt = 0.1, Gijmin = 0.92, slowing down change by 8%. When it's lower_limit = 0.0 and shares_dt = 0.05, Gijmin = 0.77. Should we explicitly set Gijmin to 1 when lower_limit (previously mes2) is zero?