Flag South Dakota partial-year model coverage instead of averaging missing policy as zero

For one adult earning $30,000/year with children ages 3 and 7 and $1,500/month per child, the app shows South Dakota subsidy of $534.90/month and ranks it near the bottom.

The model returns $1,283.75/month for August–December 2026 and zero for January–July. Its sd_cca formula explicitly says the earlier hourly county-based schedule is not implemented; the zero is a model-coverage limitation. Averaging five supported months with seven zeros produces $534.90.

This makes incomplete model coverage look like a less generous state policy, and also affects potential annual subsidy estimates.

Use a supported reference month or implement the earlier schedule upstream. Until then, visibly mark affected comparisons and annual totals as incomplete rather than presenting missing policy as an actual zero benefit. Preserve the distinction between an unmodeled period and a genuine zero payment.

Sources: scripts/precompute.py, South Dakota model.

Start with scripts/precompute.py around line 95 and compare its handling of the South Dakota period with the linked sd_cca model. Trace how supported and unsupported months flow into comparisons and annual totals. Done means incomplete coverage is visibly distinguished from a genuine zero payment, including in annual estimates.