phscs 0.2.0

Breaking changes

get_psgc() has a completely new interface. It is now re-exported directly from the psgc package. The old parameters (token, version, level, harmonize, minimal, cols) are gone. Use the new psgc parameters instead:
```
# Before
get_psgc(level = "regions")
get_psgc(grepl("^10", area_code))   # dplyr filter

# After
get_psgc(geographic_level = "region")
psgc_all <- get_psgc()
psgc_all[grepl("^10", psgc_all$psgc_code), ]   # base R subsetting
```
See vignette("psgc", package = "psgc") for full documentation.
... (dplyr filter) removed from all classification functions. get_psic(), get_psoc(), get_pcpc(), get_psced(), and get_pcoicop() no longer accept unquoted filter expressions. Filter rows with standard R subsetting after the call.
token and harmonize parameters removed from all classification functions. Data is now always sourced from the bundled local dataset — no API access or internet connection is needed.

Added get_psccs() for the Philippine Standard Commodity Classification System (PSCCS 2018).
All classification functions gain a cols = "description" option to include the full text description alongside value and label.
PCOICOP now ships both the 2020 (default) and 2009 editions. Switch with get_pcoicop(version = "2009").

Removed runtime dependencies on dplyr, tidyr, purrr, stringr, and jsonlite. The only Imports are now cli and psgc.
Minimum R version lowered from 4.1.0 to 3.5.
Data is pre-parsed at build time into a compressed sysdata.rda, making all functions substantially faster.