feat(resilience): add parse error recovery, tool timeout, and circuit breaker

4.1 Parse Error Recovery: track consecutive __parse_error tool calls;
bail with clear error after max_parse_retries (default 2) consecutive
failures. Counter resets to 0 when any tool with valid args executes.

4.2 Tool Execution Timeout: wrap execute_with_context in
tokio::time::timeout when tool_timeout_ms is set. Timeout produces an
error message fed back to the LLM so the session can recover.

4.3 Circuit Breaker: non-streaming LLM calls are retried up to
circuit_breaker_threshold times (default 3) with 100ms*attempt backoff
before the loop aborts. Streaming mode fails immediately on any error
(events cannot be replayed).

All three settings exposed via SessionOptions builder methods:
  .with_parse_retries(n)
  .with_tool_timeout(ms)
  .with_circuit_breaker(n)
  .with_resilience_defaults()  — convenience bundle

8 new tests; total: 1166 passing.

Co-Authored-By: Claude Sonnet 4.6 (1M context) <noreply@anthropic.com>

Author: ZhiXiao-Lin

README.md

@@ -11,14 +11,14 @@ let result = session.send("Refactor auth to use JWT").await?;
 [![Crates.io](https://img.shields.io/crates/v/a3s-code-core.svg)](https://crates.io/crates/a3s-code-core)
 [![Documentation](https://docs.rs/a3s-code-core/badge.svg)](https://docs.rs/a3s-code-core)
 [![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](./LICENSE)
-[![Tests](https://img.shields.io/badge/tests-1158%20passing-brightgreen.svg)](./core/tests)
+[![Tests](https://img.shields.io/badge/tests-1166%20passing-brightgreen.svg)](./core/tests)
 
 ---
 
 ## Why A3S Code?
 
 - **Embeddable** — Rust library, not a service. Node.js and Python bindings included.
-- **Production-Ready** — Permission system, HITL confirmation, skill-based tool restrictions.
+- **Production-Ready** — Permission system, HITL confirmation, skill-based tool restrictions, and error recovery (parse retries, tool timeout, circuit breaker).
 - **Extensible** — 14 trait-based extension points, all with working defaults.
 - **Scalable** — Lane-based priority queue with multi-machine task distribution.
 
@@ -476,7 +476,7 @@ cargo test          # All tests
 cargo test --lib    # Unit tests only
 ```
 
-**Test Coverage:** 1158 tests, 100% pass rate
+**Test Coverage:** 1166 tests, 100% pass rate
 
 ---