fix: 460 ruff lint errors blocking CI + Railway health check timeout

[realityinspector]

Problem 1: Ruff lint errors (CI)

The Code Quality CI step is failing with 460 ruff errors. Key issues:

• F821 Undefined names: FidelityTemporalStrategy, Timepoint (used in string type annotations in workflows/temporal_agent.py)
• F841 Unused variables: cycle_length in workflows/temporal_agent.py:674
• 460 total errors, 214 have available unsafe fixes

Run: https://github.com/timepointai/timepoint-pro/actions/runs/22907503403

Fix options

1. Add missing imports or use from __future__ import annotations to make string annotations lazy
2. Run ruff check --fix for auto-fixable issues, ruff check --fix --unsafe-fixes for the rest (review carefully)
3. If some rules are too strict for this codebase, update pyproject.toml [tool.ruff] to exclude or ignore specific rules

Problem 2: Railway health check failing

The latest Railway deployment has been stuck in INITIALIZING for 33+ hours. The /health endpoint never responds within the 30-second health check window.

Build logs show the Docker build succeeds (torch, sentence-transformers, langgraph all install fine in ~502s), but the service doesn't respond in time at startup.

Likely causes

• torch + sentence-transformers import is slow on cold start
• Missing env var causing crash before server binds
• Server binding to wrong port

Fix options

1. Check Railway env vars match what the app expects
2. Increase health check timeout in Railway service settings (try 120s)
3. Add startup logging before and after heavy imports to identify the bottleneck
4. Consider lazy-loading torch/sentence-transformers