fix(datagen): use an aligned buffer when generating random bytes arrays

[SAY-5]

RandomBytesGenerator fills a Vec<u8> and reinterprets it as T::Native through ScalarBuffer::new, but a Vec<u8> is only byte-aligned, so generating zero Float16/Decimal128/Decimal256 rows panics on the dangling pointer (#7911). Switching to MutableBuffer::from_len_zeroed gives an allocation that is aligned for any scalar type, for empty and non-empty arrays alike.

Closes #7911

[coderabbitai]

📝 Walkthrough

Walkthrough

RandomBytesGenerator now allocates aligned memory through MutableBuffer, fills it with random bytes, and wraps it for primitive arrays. A regression test covers zero-row and three-row Float16, Decimal128, and Decimal256 generation.

Changes

Random byte generation

Layer / File(s)	Summary
Aligned generation and regression coverage rust/lance-datagen/src/generator.rs	RandomBytesGenerator uses aligned zeroed storage and adds length assertions for empty and non-empty Float16 and decimal arrays.

Estimated code review effort: 2 (Simple) | ~10 minutes

Suggested reviewers: xuanwo

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title clearly matches the main change: switching random byte generation to an aligned buffer for datagen arrays.
Description check	✅ Passed	The description directly explains the alignment fix and the zero-row panic it addresses.
Linked Issues check	✅ Passed	The PR fixes #7911 by using aligned allocation and adds coverage for empty Float16 and decimal arrays.
Out of Scope Changes check	✅ Passed	The changes stay focused on the alignment bug fix and its regression test, with no clear unrelated additions.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

_{Comment @coderabbitai help to get the list of available commands.}

[coderabbitai]

🧹 Nitpick comments (1)

rust/lance-datagen/src/generator.rs (1)

3532-3548: 🎯 Functional Correctness | 🔵 Trivial | ⚡ Quick win

Exercise typed value access in the regression test.

The test currently verifies only Array::len(). Add assertions for the expected data_type() and concrete primitive array downcasts, then access the non-empty values so the test proves alignment remains valid beyond construction.

As per coding guidelines, every bugfix and feature must have corresponding tests.

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@rust/lance-datagen/src/generator.rs` around lines 3532 - 3548, Extend
test_rand_bytes_zero_rows to validate each generated array’s expected data_type
and downcast it to the corresponding concrete primitive array type, then access
the values from the three-row result. Retain the existing zero- and
non-empty-length assertions so the test covers both construction and typed value
access for Float16, Decimal128, and Decimal256.

Source: Coding guidelines

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Nitpick comments:
In `@rust/lance-datagen/src/generator.rs`:
- Around line 3532-3548: Extend test_rand_bytes_zero_rows to validate each
generated array’s expected data_type and downcast it to the corresponding
concrete primitive array type, then access the values from the three-row result.
Retain the existing zero- and non-empty-length assertions so the test covers
both construction and typed value access for Float16, Decimal128, and
Decimal256.

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: QUIET

Plan: Pro Plus

Run ID: 0782fe02-1d46-4d35-9166-fb880eb1264a

📥 Commits

Reviewing files that changed from the base of the PR and between be778df and 04f44e3.

📒 Files selected for processing (1)

• rust/lance-datagen/src/generator.rs