Add support for structured datatypes

[Bisaloo]

No description provided.

[codecov-commenter]

⚠️ Please install the to ensure uploads and comments are reliably processed by Codecov.

Codecov Report

❌ Patch coverage is 95.76271% with 5 lines in your changes missing coverage. Please review.
✅ Project coverage is 85.51%. Comparing base (a827ae9) to head (1a54ed6).

Files with missing lines	Patch %	Lines
R/read_npy.R	95.76%	5 Missing ⚠️
❗ Your organization needs to install the Codecov GitHub app to enable full functionality.

Additional details and impacted files

@@            Coverage Diff             @@
##             main       #3      +/-   ##
==========================================
- Coverage   85.71%   85.51%   -0.20%     
==========================================
  Files           6        6              
  Lines         343      366      +23     
==========================================
+ Hits          294      313      +19     
- Misses         49       53       +4     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[github-actions]

This is how benchmark results would change (along with a 95% confidence interval in relative change) if 1a54ed6 is merged into main:

• ❗🐌pkg_load: 4.77ms -> 4.9ms [+0.38%, +5.18%]
• ❗🐌read_bigendian: 1.14ms -> 11.3ms [+888.76%, +900.59%]
• ❗🐌read_bool: 1.15ms -> 11.5ms [+891.12%, +904.39%]
• ❗🐌read_empty: 1.09ms -> 11ms [+909.82%, +914.02%]
• ❗🐌read_float32: 1.19ms -> 12.1ms [+895.23%, +928.48%]
• ❗🐌read_float64: 1.18ms -> 12ms [+898.02%, +931.67%]
• ❗🐌read_int16: 1.17ms -> 11.9ms [+908.83%, +921.17%]
• ❗🐌read_int32: 1.2ms -> 11.8ms [+875.94%, +891.64%]
• ❗🐌read_int8: 1.17ms -> 11.8ms [+902.84%, +916.49%]
• ❗🐌read_npz: 2.81ms -> 12.8ms [+354.67%, +358.35%]
• ❗🐌read_string: 1.13ms -> 11.4ms [+901.83%, +919.91%]
• ❗🐌read_uint16: 1.09ms -> 11.1ms [+912.75%, +924.11%]
• ❗🐌read_uint32: 1.4ms -> 11.8ms [+730.44%, +745.43%]
• ❗🐌read_uint64: 1.53ms -> 11.6ms [+653.77%, +664.16%]
• ❗🐌read_uint8: 1.09ms -> 11.2ms [+914.96%, +926.92%]
• ❗🐌read_unicode: 1.19ms -> 11.1ms [+833.55%, +838.63%]
  Further explanation regarding interpretation and methodology can be found in the documentation.

[Bisaloo]

Slowdown only impacts header parsing, which means it's not noticeable in larger arrays. In smaller arrays, it doesn't matter:

cross::bench_branches(
  {
    library(grumpy)
    bench::mark(
      read_npy(system.file("extdata", "test.npy", package = "grumpy")),
      iterations = 50
    )
  }
)

✔ Installing branch 'structured-type'
✔ Installing branch 'main'
✔ Running `expr` across variants
# A tibble: 2 × 14
  branch          expression                   min median `itr/sec` mem_alloc `gc/sec` n_itr  n_gc total_time result   memory     time       gc      
  <chr>           <bch:expr>                 <bch> <bch:>     <dbl> <bch:byt>    <dbl> <int> <dbl>   <bch:tm> <list>   <list>     <list>     <list>  
1 structured-type "read_npy(system.file(\"e… 212µs  251µs     3959.    1.57MB     80.8    49     1     12.4ms <dbl[…]> <Rprofmem> <bench_tm> <tibble>
2 main            "read_npy(system.file(\"e… 253µs  284µs     3462.  437.27KB     70.6    49     1     14.2ms <dbl[…]> <Rprofmem> <bench_tm> <tibble>