Add Laya request preprocessing - #43

Purpose

Pack English Laya requests into token rows, option-marker positions, question types and usage counts. Preserve question and option order and apply Laya 0.3.20's truncation rules.

#21 is merged. This PR adds preprocessing on top of that checkpoint layer: 496 core/configuration lines, excluding separate tests, docs and Cargo.lock. Refs #14.

Also preserve Cua-S1 JSON number parsing and Python float formatting when Laya enables serde_json/arbitrary_precision. Without this fix, three existing Cua-S1 tests fail in the combined workspace.

Fix private Number/RawValue key collisions in structured Laya inputs. Native callers now use Request::from_json(&str) or Request::from_value(Value); generic Deserialize is removed, while public fields and Serialize remain. New tests and helpers, including the Cua-S1 regression, live under repository-root tests/.

Test Plan

Run workspace tests, fmt, strict Clippy and a release build. Compare packed inputs against the frozen 17-case Laya reference. Check private object keys, numeric preservation, request fields and JSON/value depth boundaries. Pinned inputs and commands are in the CPU packing recipe.

System1-Omni Version / Commit: 37736f0. Original preprocessing increment: 219808b → 99c1d12.

Test Result

• Local workspace and current CPU CI: 39 tests passed and 6 external-data/GPU tests were skipped. The official 17-case packing comparison passed separately; private-key regression tests failed before the fix and passed after it.
• Earlier Cua-S1 checks passed all 7 unit tests both with and without arbitrary_precision; the production parser is unchanged by the test relocation.
• fmt, strict Clippy, release build and strict MkDocs build passed locally.
• Earlier full-checkpoint checks passed; checkpoint code is unchanged. No GPU inference or model-quality tests were run for this update.

CI for 37736f0: Rust CI, Docs build, benchmark harness tests passed.

Self-review

Before marking this PR ready for review or requesting maintainer review, complete
the self-review checklist.
Keep the PR in draft while this work is incomplete.
For agent assistance, use the optional precheck-pr skill.

• I have reviewed the full diff and addressed the issues I found.
• I have checked that the change follows the project's architecture and stays focused on the stated purpose.
• I have run the checks appropriate to this change and reported commands, results, and anything I could not verify above.
• I have checked that the PR description, documentation, and any accuracy or performance claims match the implementation and available evidence.