Derive PUMS recode_years from pums_variables (remove hard-coding)

[ashleyasmus]

What this does

Closes #634

load_data.R currently hard-codes the year range used to determine whether PUMS recoding is supported (via recode_years <- ...). This PR derives the supported years directly from the packaged pums_variables dataset (scoped to the requested survey), so adding new PUMS dictionaries doesn’t require updating code in load_data.R.

No intended behavior changes beyond removing the manual maintenance step.

Why

pums_variables already contains the authoritative list of survey/year combinations with dictionaries (and thus recode support). Using it eliminates the need to update recode_years annually.

Tests I ran

• 2023 acs1 recode works (labels returned):

get_pums(state = "Vermont", variables = c("AGEP","SEX"), year = 2023, survey = "acs1", recode = TRUE)

Result includes STATE_label and SEX_label columns.

• 2024 acs1 currently does not recode on main (expected until dictionary update):

get_pums(state = "Vermont", variables = c("AGEP","SEX"), year = 2024, survey = "acs1", recode = TRUE)

Result shows the current message:

Recoding is currently supported for 2017 - 2023 data. Returning original data only.

(With this PR, the supported-year check is no longer hard-coded and will automatically reflect the years present in pums_variables once 2024 is added.)

Notes

• This PR only changes how supported recode years are determined; it does not add or modify any PUMS dictionaries/data objects.
• Companion PR Add 2024 ACS 1-year PUMS data dictionary + enable 2024 recoding #635 adds 2024 acs1 dictionary support.