try pkg-config before gdal-config

[rsbivand]

In relation to @s-u s email today, an attempt at trying pkg-config before falling back on gdal-config. Also small change to test for stateplane.csv not pcs.csv which disappeared some time ago in the GDAL data folder.

[rsbivand]

Just added test for CURL missing from GDAL build. When run locally, it gives:

> aa[[114]]
$file
[1] "test-read.R"

$context
NULL

$test
[1] "GDAL was built with CURL"

$user
[1] 0.198

$system
[1] 0.001

$real
[1] 0.7

$results
$results[[1]]
<expectation_success/expectation/condition>
As expected

With macOS static GDAL 3.10.1:

── Error (test-read.R:211:2): GDAL was built with CURL ─────────────────────────
<Rcpp::exception/C++Error/error/condition>
Error: Cannot open "https://github.com/r-spatial/sf/raw/refs/heads/main/inst/gpkg/nc.gpkg"; The file doesn't seem to exist.
Backtrace:
     ▆
  1. ├─testthat::expect_silent(x <- read_sf("https://github.com/r-spatial/sf/raw/refs/heads/main/inst/gpkg/nc.gpkg")) at test-read.R:211:9
  2. │ └─testthat:::quasi_capture(enquo(object), NULL, evaluate_promise)
  3. │   ├─testthat (local) .capture(...)
  4. │   │ ├─withr::with_output_sink(...)
  5. │   │ │ └─base::force(code)
  6. │   │ ├─base::withCallingHandlers(...)
  7. │   │ └─base::withVisible(code)
  8. │   └─rlang::eval_bare(quo_get_expr(.quo), quo_get_env(.quo))
  9. └─sf::read_sf("https://github.com/r-spatial/sf/raw/refs/heads/main/inst/gpkg/nc.gpkg")
 10.   ├─sf::st_read(...)
 11.   └─sf:::st_read.character(...)
 12.     └─sf:::CPL_read_ogr(...)

[ FAIL 1 | WARN 3 | SKIP 67 | PASS 734 ]
Error: Test failures

[rsbivand]

@edzer, @s-u, with the proposed test, Simon would have had a chance of catching the regression when checking sf. Maybe we will not fall down the same manhole again with it?