Uprate SPM thresholds with a more realistic index

[MaxGhenis]

The Census Bureau defines SPM thresholds based on expenditures. This rises consistently faster than CPI, which is what we currently index to:

https://github.com/PolicyEngine/policyengine-us/blob/6525e72ffc89aca67e88ba6381dfa063be0123bc/policyengine_us/variables/household/income/spm_unit/spm_unit_spm_threshold.py#L10

Let's either identify a forecast of expenditures (ideally on the relevant categories: food, clothing, shelter, and utilities), or otherwise construct one based on CPI plus the typical delta.

[MaxGhenis]

Census provided this comparison of CPI-U to FCSUti (which governs SPM threshold growth): https://www.bls.gov/pir/spm/spm_thresholds_2023.htm#_edn1

Since 2010, CPI-growth has averaged 2.55%, while FCSUti has averaged 2.71% (average annual rate, not CAGR).

However, the five-year moving average nature of FCSUti, combined with rapid growth in 2021 and 2022, means that the faster growth is likely to continue for the next couple years. I'd suggest then that we take these steps:

1. Collect the annual levels of FCSU
2. Estimate the annual growth of FCSU, compared to CPI-U (e.g. 0.2pp above CPI-U over the past 10 years)
3. Project each year of FCSU based on (2) above the current CPI-U forecast (for now - would ideally forecast price growth for these categories specifically)
4. Calculate each year's SPM threshold based on the appropriate 5-year moving average
5. Write this to a yaml file to uprate thresholds by

We could do this as a spreadsheet for now, though ideally it would be a Python script.