feat(tools): add standardized aden_tools logging

[Jonathan1366]

Summary

aden_tools currently does not have a shared logging standard. Errors are correctly returned to callers, but maintainers don’t have consistent runtime logs to help with debugging and operational visibility—especially important for MCP STDIO, where stdout must remain clean for JSON-RPC.

We also ran into a practical issue during development: Python logging can raise a runtime KeyError when extra contains reserved LogRecord keys (for example extra={"name": ...}), which can unexpectedly crash otherwise normal error paths.

---

Proposal

• Add a small, centralized logging helper:

  • configure_logging() (safe to call multiple times)
  • get_logger()

• Default all logs to stderr (STDIO / JSON-RPC safe).

• Allow basic configuration via environment variables:

  • ADEN_TOOLS_LOG_LEVEL (default: INFO)
  • ADEN_TOOLS_LOG_FORMAT

• Standardize safe structured log metadata:

  • Avoid reserved LogRecord fields such as name, message, and asctime.

• Add a lightweight test to ensure:

  • Logging initialization is idempotent (no duplicate handlers).

---

Example (real issue → safe pattern)

Before

logger.warning(
    "pdf_read file not found",
    extra={"name": path.name}
)

This can raise a runtime KeyError due to LogRecord field collisions.

After

logger.warning(
    "pdf_read file not found",
    extra={"file_name": path.name}
)

Uses a neutral metadata key and avoids collisions.

---

Test plan

cd tools
python3 -m pip install -e ".[dev]"
python3 -m pytest tests

---

[Jonathan1366]

I'd like to work on this! Please assign for me, thanks @TimothyZhang7 @vincentjiang777 @bryanadenhq

Proposed approach:

• Add aden_tools.utils.logging with configure_logging() + get_logger() (idempotent, stderr by default, env-configurable via ADEN_TOOLS_LOG_LEVEL / ADEN_TOOLS_LOG_FORMAT).
• Use the logger in tools/mcp_server.py and a few core tools (minimal metadata only; no secrets).
• Avoid Python LogRecord extra collisions (do not use reserved keys like name; use file_name, status_code, etc.).
• Add a small unit test to guarantee idempotent setup (no duplicate handlers).

If this direction looks good, I’ll open a PR with the implementation + tests (cd tools && pytest tests).