feat(array): cast bool and primitive values to Utf8

[haohuaijin]

Rationale for this change

DataFusion can push CAST(... AS STRING) into Vortex scans, but boolean and primitive arrays currently lack native casts to Utf8, causing pushed queries to fail with a missing CastKernel.

Related to #6211, #8621, and #6702.

What changes are included in this PR?

• Add BoolArray -> Utf8 and PrimitiveArray -> Utf8 casts.
• Add matching scalar casts for constant and nested values. This includes Bool -> Primitive so they behave consistently with the existing BoolArray -> Primitive cast (true = 1, false = 0).
• Preserve nullability and use Arrow-compatible numeric formatting.
• Add unit tests and a DataFusion projection-pushdown E2E test.

Validation:

• cargo nextest run -p vortex-array
• cargo nextest run -p vortex-datafusion
  • Includes an E2E test that writes nullable Boolean, Int64, and Float64 columns to a Vortex file, runs CAST(... AS STRING) with projection pushdown enabled, and verifies both values and nulls.
  • The E2E test fails on develop with a missing Bool -> Utf8 cast kernel and passes with this PR.
• cargo clippy -p vortex-datafusion --all-targets --all-features
• cargo +nightly fmt --all

What APIs are changed? Are there any user-facing changes?

No public API signatures are changed.

Boolean and primitive values can now be cast to Utf8 inside Vortex, including DataFusion projection pushdown.

[haohuaijin]

hi @AdamGS , do you have time take a look?

[codspeed-hq]

Merging this PR will degrade performance by 2.12%

⚡ 2 improved benchmarks
❌ 2 regressed benchmarks
✅ 1958 untouched benchmarks
⏩ 89 skipped benchmarks¹

Warning

Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
❌	Simulation	search_index_in_range_chunked	5.4 ms	6.4 ms	-15.17%
❌	Simulation	search_index_above_max_chunked	4.7 ms	5.4 ms	-12.33%
⚡	Simulation	chunked_dict_primitive_canonical_into[u32, (1000, 1000, 100)]	2 ms	1.8 ms	+11.33%
⚡	Simulation	chunked_dict_primitive_canonical_into[u32, (1000, 10, 100)]	1.6 ms	1.4 ms	+10.84%

Tip

Investigate this regression by commenting @codspeedbot fix this regression on this PR, or directly use the CodSpeed MCP with your agent.

---

_{Comparing haohuaijin:codex/bool-primitive-to-utf8 (2c3825c) with develop (c4bb934)}

Footnotes

1. 89 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩

[joseph-isaacs]

please use generics and a macro each macro (we use elsewhere)

[joseph-isaacs]

why did you need the new trait where did the other definitions come from where could we place this code?

[haohuaijin]

this is for unify the float to utf8 convert to use append_float_values_to_utf8 func

fn append_float_values_to_utf8<T>(builder: &mut VarBinViewBuilder, values: &[T], mask: &Mask)
where
    T: FloatToUtf8,
{
    let mut formatter = T::formatter();
    append_values_to_utf8(builder, values, mask, |builder, value| {
        builder.append_value(value.format(&mut formatter));
    });
}

i follower the arrow-rs, float32/float64 use ryu, float16 use display, https://github.com/apache/arrow-rs/blob/fc16607ccfdb298557ed6630742941761ac18ce3/arrow-cast/src/display.rs#L711-L734

[joseph-isaacs]

Why did you add bool -> prim here too?

[haohuaijin]

Thanks for your reviews @joseph-isaacs , i apply you suggestion in 2c3825c.

Why did you add bool -> prim here too?

sorry make you misunderstand, the pr #8621 impl the bool -> prim cast, but only for BoolArray, so in this pr i also add it for BoolScalar. should i remove it?

[joseph-isaacs]

can you explain the trick

[haohuaijin]

first init the full buffer with more frequent value(true/false), then only overwrite the another(false/true) positions, that minimize the index write.

[joseph-isaacs]

This new a few comments explain the complex/interesting stuff and thoughts about where to put new trait or can we extend existing ones?

[joseph-isaacs]

please do split

[haohuaijin]

please do split

split it out of this pr