Stabilize MCP refresh regression tests

[henrypark133]

Summary

• make the MCP proxy-refresh regression tests deterministic by isolating them from shared OAuth proxy env state
• use a local token URL in the proxy-configured MCP refresh test and remove allocator-sensitive lock pointer assertions
• align the proxy refresh mock response with the token metadata assertions for token_type and scope

Testing

• env CARGO_TARGET_DIR=/tmp/ironclaw-postmerge-test cargo test --no-default-features --features libsql test_refresh_ --lib

[Copilot]

Pull request overview

Stabilizes MCP OAuth proxy refresh regression tests by removing dependency on shared process env state and aligning mocked proxy responses with token metadata expectations.

Changes:

• Make proxy-refresh regression tests deterministic by guarding and resetting OAuth proxy-related env vars during tests.
• Use a local, dynamically generated token_url for the proxy-configured MCP refresh test.
• Replace allocator-sensitive lock pointer assertions with Weak/Arc::ptr_eq-based assertions and align mock proxy responses to include token_type and scope.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated no comments.

File	Description
src/tools/mcp/auth.rs	Makes MCP refresh/proxy tests deterministic (env isolation, local token URL) and makes refresh-lock tests robust by avoiding pointer-based assertions.
src/cli/oauth_defaults.rs	Updates mock proxy token responses to include token_type and scope to match token metadata assertions.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[gemini-code-assist]

Code Review

This pull request enhances the test suite for OAuth and MCP authentication. It updates mock JSON responses to include token_type and scope, replaces hardcoded test URLs with dynamic values, and improves concurrency tests by implementing environment variable guards and more precise Arc pointer comparisons. I have no feedback to provide.