Take consumption into account on attention days

[sbenthall]

Currently, on an agent's attention day, they compute their share demand by running their Share function on the normalized assets value aNrm.

https://github.com/sbenthall/HARK_ABM_INTRO_public/blob/master/HARK/hark_portfolio_agents.py#L960-L995

But this not a correct use of the Share function, which takes 'cash on hand', mNrm, as an input normally. The Share function takes into account the expectation that some of these resources will be consumed before choosing to invest in the risky asset.

econ-ark/HARK#1064

So, running this function on the aNrm is going to give an incorrect Share value, because it will assume that more of the aNrm have been consumed already. This is not what's going on in the simulation -- the agents will already have consumed on the prior macro day.

We can correct for this by running the Share function on the reconstructed cash on hand, by "giving back" what they consumed in the prior macro day.

asset_normalized = agent.state_now['aNrm'] + agent.control['cNrm']

[sbenthall]

Need to use sequential portfolio type, now in HARK