fix: inconsistency inline constant handling for non‑fixed‑stride datatype

[zhangyue19921010]

Encoding side (rust/lance-arrow/src/scalar.rs)

try_inline_value decides whether a scalar value can be written inline. The old logic only checked that:

• the scalar is non-null and a single value
• it has no child data
• there is only one buffer
• the inline size does not exceed the threshold

However, it did not verify that the data type is fixed-stride or not. TAKE BOOLEAN AS EXAMPLE

Decoding side (decode_scalar_from_inline_value)

Decoding explicitly requires data_type.byte_width_opt() to be present; otherwise it errors out:

• Inline constant is not supported for non-fixed-stride data type Boolean.

This creates an inconsistency between the encoder and decoder.

Without this fix, new added unit test test_constant_layout_out_of_line_boolean_v2_2 and similar scenarios will report errors

called `Result::unwrap()` on an `Err` value: Arrow { message: "Invalid argument error: 
Inline constant is not supported for non-fixed-stride data type Boolean", 
location: Location { file: "rust/lance-encoding/src/encodings/logical/primitive/constant.rs", line: 246, column: 59 } }

[zhangyue19921010]

Hi @Xuanwo , would u mind to take a look at your convenient? Thanks!

[codecov]

Codecov Report

❌ Patch coverage is 91.30435% with 2 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
.../lance-encoding/src/encodings/logical/primitive.rs	89.47%	2 Missing ⚠️

📢 Thoughts on this report? Let us know!

[wjones127]

This seems reasonable, but would like @Xuanwo to take a look before we merge.

[zhangyue19921010]

Thanks for your help @wjones127 changed.

[westonpace]

One minor nit and then we can merge. Just ping me when you've addressed it.

[westonpace]

Can we remove the with_max_file_version? There is no reason this test should stop working in future versions.

[zhangyue19921010]

This Bug is fixed by another PR #5862

Thanks for your attention and help here @westonpace and @wjones127 ：）