perf: cover signed by in core benchmark

[tiye]

Summary

• add signed-by-iteration to the standard cargo bench --bench core suite
• keep a runtime semantic guard for expected value 3333
• record a core-benchmark sample alongside the qbench sample in PERFORMANCE.md

Validation

• cargo bench --locked --bench core
• make check
• git diff --check

This PR is chained after PR #39 and is intentionally not merged automatically.

[Copilot]

Pull request overview

Adds the signed-step workload to the core benchmark suite and records its timing sample.

Changes:

• Adds signed-by-iteration with expected value 3333.
• Documents core benchmark results in PERFORMANCE.md.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 1 comment.

File	Summary	Review comment
PERFORMANCE.md	Records benchmark timing data.	Nit: Align the documented iteration counts for consistency.
benches/core.rs	Adds the signed-step workload and semantic guard.	—

Suppressed comments (1)

PERFORMANCE.md:548

• This adds only a single core-benchmark reading, while the report's current baseline/raw-sample sections still omit signed-by-iteration and explicitly require three release runs with a median for benchmark evidence (PERFORMANCE.md:493-534; RFCs/0045-semantic-benchmark-guards.md:6-8). Please either add this workload to the same three-run median tables, or clearly label this as a non-baseline validation sample like the preceding qbench entry so it is not presented as comparable core-benchmark data.

标准 `cargo bench --locked --bench core`（10,000 次）同样已纳入该负载；本次样本为编译 `38.544 ms`、验证 `2.032 ms`、执行 `157.776 ms`，期望值 `3333`。它与正向 `stepped-iteration` 的执行样本（`124.616 ms`）同场输出，便于观察有符号步进的额外边界检查成本。

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.