Create raft of tests for structural reforms

There would be a lot of value to adding a series of tests around creating structural reforms with the following:

• A reform using a neutralize_variable value on an existing piece of the default tax-benefit system
• A reform using an update_variable value on an existing piece of the default tax-benefit system
• A reform using an update_variable value to create a totally new variable in the system

All of these should be run in the following four-step manner:

• For a relevant test household, baseline, no axes
• Test household, reform, no axes
• Test household, baseline, axes
• Test household, reform, axes

This mimics the series of calls that occurs if a user inputs a reform, inputs a household, views their net income output, then clicks "Varying your earnings".

Each output should confirm that:

• household_net_income is a valid value
• Each of the default outputs is a valid value (we could even drill down the tree using metadata to ensure every possible display value is valid, depending on how long that makes the test time-wise)
• Some variable added or altered by the structural reform is present and valid