Propagate RLM stop errors from root and sub tools

[snimu]

Description

Make sure all tools (root_tools and sub_tools and thus also tools) propagate stop_errors to call_bash/python_repl, so that StatefulToolEnv's handling of stop_error still works.

Type of Change

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to not work as expected)
• Documentation update
• Test improvement

Testing

• All existing tests pass when running uv run pytest locally.
• New tests have been added to cover the changes

Checklist

• My code follows the style guidelines of this project as outlined in AGENTS.md
• I have performed a self-review of my own code
• I have commented my code, particularly in hard-to-understand areas
• I have made corresponding changes to the documentation
• My changes generate no new warnings
• Any dependent changes have been merged and published

---

Note

Ensures RLM stop errors raised by tools are surfaced to the REPL so StatefulToolEnv stop conditions trigger correctly.

• In sub-tool execution and llm_batch batching, re-raise exceptions that satisfy _should_stop_for_error, instead of swallowing them as error messages
• In interception handlers for root tools and sub-LLM requests, set state_ref["_rlm_stop_error"] when a stop error is detected
• In REPL execution (_call_repl), pop and re-raise _rlm_stop_error after code execution to propagate the stop upstream

^{Written by Cursor Bugbot for commit 3c92ad5. This will update automatically on new commits. Configure here.}

[cursor]

Cursor Bugbot has reviewed your changes and found 1 potential issue.

^{Bugbot Autofix is OFF. To automatically fix reported issues with Cloud Agents, enable Autofix in the Cursor dashboard.}