iterative algorithm for take_value

[davidhewitt]

Similar to #111, but for real work in take_value. If this adds a perf boost I can apply the same design to the Python functions too...

[codecov]

Codecov Report

Attention: Patch coverage is 95.18072% with 8 lines in your changes missing coverage. Please review.

Project coverage is 94.01%. Comparing base (dd25fd0) to head (08b4cd3).

Files with missing lines	Patch %	Lines
crates/jiter/src/value.rs	95.18%	8 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #134      +/-   ##
==========================================
- Coverage   94.91%   94.01%   -0.91%     
==========================================
  Files          12       12              
  Lines        1908     2038     +130     
==========================================
+ Hits         1811     1916     +105     
- Misses         97      122      +25     

Files with missing lines	Coverage Δ
crates/jiter/src/lazy_index_map.rs	85.22% <ø> (-12.76%)	⬇️
crates/jiter/src/value.rs	94.26% <95.18%> (-2.85%)	⬇️

---

Continue to review full report in Codecov by Sentry.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update dd25fd0...08b4cd3. Read the comment docs.

[codspeed-hq]

CodSpeed Performance Report

Merging #134 will degrade performances by 10.97%

_{Comparing dh/iterative-take (546c4df) with main (dd25fd0)}

Summary

⚡ 1 improvements
❌ 1 regressions
✅ 71 untouched benchmarks

⚠️ Please fix the performance issues or acknowledge them on CodSpeed.

Benchmarks breakdown

	Benchmark	main	dh/iterative-take	Change
⚡	pass2_jiter_value	22.1 µs	17.1 µs	+29.51%
❌	x100_jiter_skip	1.2 µs	1.3 µs	-10.97%

[davidhewitt]

The slowdown is probably an inlining loss in unrelated code which can be recovered by PGO. The +24% seems very real.

[samuelcolvin]

Lgtm