["model"] Fitting won't match to post-peak dates

### Summary

The fix for #378 forces the fitting procedure to match up the census to the closest date *pre-peak*, so that we don't accidentally pick a day in a long projection that's way past the peak instead.

### Additional details

### Suggested fix

Start by reading the fix for #378 and locating the fitting procedure that matches census dates to the model peak. Confirm how post-peak dates are currently handled, then verify that matching stays on the pre-peak side without changing the model's validated behavior.

Scritto dal modello di indicizzazione a partire dal testo della issue.