feat(web): add Qwen native search adapter

[h3c-hexin]

Summary

• Enable provider-native web search for qwen3.8-max, qwen3.7-plus, and qwen3.7-max on ModelStudio Token Plan's Responses Harness.
• Send the documented web_search tool with string tool_choice: "required" through the active authenticated transport.
• Keep preview, Coding Plan, Anthropic, and custom-compatible routes fail-closed.

Refs #5681 (partial).

No-Issue: partial delivery tracked by #5681; this slice must not close the umbrella issue.
Related: #5682.

Testing

• cargo fmt --all -- --check
• cargo test -p codewhale-config qwen_native_search_is_exact_to_token_plan_responses_routes -- --nocapture
• cargo test -p codewhale-tui client::provider_native_search::tests -- --nocapture — 9 passed
• cargo check -p codewhale-config -p codewhale-tui
• git diff --check

Checklist

• Exact product/model gate with negative adjacent-route coverage
• Request and response contracts covered by deterministic HTTP mocks
• Provider documentation and changelog updated
• No credentials, UI changes, or unrelated provider adapters

Honesty boundary

This exact upstream commit was verified with deterministic mocks, not a funded ModelStudio Token Plan request. No credentials, release, tag, deployment, or publication action is part of this PR. The workspace-wide clippy and test baseline findings are already recorded on #5682; this slice's changed config and provider-native test targets pass.

[Hmbown]

Merged via current-main rescue PR #5690 (merge commit ) — direct pushes to the fork branch were denied (403), so the reconciliation merge carrying your exact head as second parent landed through #5690. All three commits and your authorship are preserved verbatim in main. Thank you for the contribution and for fixing the changelog/link gates on the original branch.