docs(arrow-convert): document and doctest the ArrowData read-back API

[phil-opp]

Issue

ArrowData's TryFrom<&ArrowData> implementations (libraries/arrow-convert/src/from_impls.rs) are the primary public API for reading received Arrow data back into Rust values — the read-back counterpart to IntoArrow. IntoArrow has a documented, compile-checked doctest (lib.rs), but the read-back side had none, and its non-obvious, load-bearing contract was only discoverable by reading the source:

• scalar conversions (u8, f32, String, &str, chrono types) require exactly one element and no nulls;
• slice / Vec conversions accept any length but reject any nulls.

Fix

Expand the ArrowData doc comment to describe both conversion shapes and their length/null contracts, with a compile-checked doctest that:

• round-trips a scalar (u8), a String, and a Vec<i32> via try_into(), and
• shows that a multi-element array cannot be read as a scalar (returns Err).

Documentation only — no behavior change. This is a good fit for the "prefer compile-checked doctests" documentation goal: the doctest locks the contract against future drift.

Validation

• cargo test -p dora-arrow-convert --doc — 2 passed (the new ArrowData doctest plus the existing IntoArrow one).
• cargo fmt --all -- --check and cargo clippy -p dora-arrow-convert -- -D warnings clean.

---

🤖 This is a machine-generated pull request opened by Claude Code as part of an automated code-review pass. Please review carefully before merging.

🤖 Generated with Claude Code

---

Generated by Claude Code

[trunk-io]

😎 Merged successfully - details.

[phil-opp]

🤖 Automated review by Claude — this is a fully automated review with no human in the loop.

I reviewed this diff and found no issues. It is a docs-only change adding a doc comment plus a compile-checked doctest for ArrowData's TryFrom<&ArrowData> read-back API; no behavior change. The doctest correctly exercises the scalar single-element rule and the Vec/any-length rule, and asserts that a multi-element array cannot be read back as a scalar.

---

Generated by Claude Code

[github-actions]

@phil-opp the Trunk merge queue failed for this PR.

See the Trunk merge-status comment for details.

Posted as a new comment so GitHub sends an email — Trunk's sticky comment is edited in place and won't trigger a notification.

[phil-opp]

These fault_tolerance_tests failures don't originate from this PR. #3084 only adds documentation and a doctest to libraries/extensions/arrow-convert — it doesn't touch binaries/daemon.

The failing Test (ubuntu-latest) ran against the Trunk merge-queue batch branch (533dc2a), which bundles this PR together with others ahead of it in the queue. The panic locations (binaries/daemon/src/lib.rs:8289 / 8327 / 8377 / 8407) sit ~20 lines below where these tests actually live on main — e.g. wedged_user_input_node_is_a_finish_straggler's assert is at lib.rs:8268 here — so the batch contains another PR that shifted the file and changed finish_stragglers classification behavior.

Verified locally on current main: all four named tests, and the full fault_tolerance_tests suite (55 tests), pass:

test result: ok. 55 passed; 0 failed; 0 ignored; 0 measured; 149 filtered out

Nothing to fix on this PR — it should re-test clean once the queue re-batches it without the offending sibling.

---

Generated by Claude Code