Add ArrayPlugin to enable serde for new DecimalBytePartsArray format

[mhk197]

Add serialization for DecimalBytePartsArray with lower parts while preserving the frozen legacy format (no lower parts). The in-memory DecimalBytePartsArray can now use either wire format, selected from its child layout:

Array layout	Serialized ID
Signed most significant part (MSP) only	vortex.decimal_byte_parts
Signed MSP plus one to three unsigned 64-bit lower parts	vortex.decimal_byte_parts_v2

Both formats deserialize into the same array representation. The metadata records the MSP's physical integer type and the lower-part count. Decoding preserves valid physical storage that is wider than the minimum required by the decimal's precision.

DecimalBytePartsPlugin registers both serialized IDs and selects the appropriate one when writing. On read, it enforces each ID's contract: the frozen format must have no lower parts, and v2 must have at least one.

Both vtable serde and plugin serde remain supported:

• Vtable serde preserves the existing API. Directly registering DecimalByteParts continues to read and write the frozen format. It rejects lower parts, so existing callers cannot accidentally write a wider layout under the old ID. Keeping this path avoids breaking callers that use direct vtable registration or its serde methods.
• Plugin serde adds format selection. Registering DecimalBytePartsPlugin supports both formats. The plugin can return a distinct serialized ID, which the vtable's metadata-only serializer cannot do. vortex_decimal_byte_parts::initialize registers this plugin.

Add a separate v2 compatibility fixture covering positive and negative wide i128 values and nullable i256 values. The existing frozen fixture is unchanged.

[codspeed-hq]

Merging this PR will improve performance by 11.7%

⚠️ Unknown Walltime execution environment detected

Using the Walltime instrument on standard Hosted Runners will lead to inconsistent data.

For the most accurate results, we recommend using CodSpeed Macro Runners: bare-metal machines fine-tuned for performance measurement consistency.

⚠️ Different runtime environments detected

Some benchmarks with significant performance changes were compared across different runtime environments,
which may affect the accuracy of the results.

Open the report in CodSpeed to investigate

⚡ 3 improved benchmarks
✅ 2202 untouched benchmarks
⏩ 224 skipped benchmarks¹

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
⚡	Simulation	allocate_drop_arrow[0]	456.9 ns	402.7 ns	+13.45%
⚡	WallTime	mul_u32_nonnull_avx512	6.3 µs	5.6 µs	+11.29%
⚡	Simulation	allocate_drop_bytes[0]	575.7 ns	521.6 ns	+10.39%

Tip

Curious why performance improved? Comment @codspeedbot explain why performance improved on this PR, or directly use the CodSpeed MCP with your agent.

---

_{Comparing mk/dbp-serde (03a13f8) with mk/dbp-array (7283d4f)}

Footnotes

1. 224 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. ↩