Fix North Carolina monthly copay being divided by 12 twice

For one working adult earning $30,000/year with children ages 3 and 7, the default $1,500/month charge per child, and the default North Carolina provider/county, the app displays a $20.83/month copay (rounded to $21). The underlying model returns $250 for each month.

Reproduced against the committed grid (policyengine-us 1.824.8):

• nc_scca_parent_fee requested for 2026: 250.
• The same variable requested for each of the 12 months: 250 each, totaling 3,000.
• monthly_copay() divides the annual-request value by 12, incorrectly producing 20.83.

An all-state check for this reference household found NC was the only copay whose annual request differed from the sum of its monthly requests. This is a period-aggregation issue, not a claim that the state's fee rate is wrong.

Calculate monthly values explicitly and average them, preserving daily/weekly conversion factors. Audit the live-result reader, which makes the same annual-sum assumption; regenerate the affected state and comparison data. Add a regression using the actual model's monthly versus annual response behavior.

Sources: scripts/precompute.py, src/lib/api.ts.

Start by reading scripts/precompute.py and src/lib/api.ts, then reproduce the North Carolina household case against policyengine-us 1.824.8. Trace both precomputed and live-result period aggregation, add the regression for the model's monthly versus annual responses, and confirm affected state and comparison data are regenerated with daily and weekly conversions preserved.