feat: kbagent suggest — contextual recommendations for AI agents

[padak]

Problem

`kbagent context` is a static document. Once an agent reads it, it works from memory. But agents often need contextual guidance:

• "I see 3 failing jobs — what should I investigate first?"
• "Is there a faster way to do what I'm about to do?"
• "What are the dependencies I should check before modifying this config?"

Proposal

kbagent suggest --project prod --goal "debug failing jobs"

Output: prioritized, contextual action plan based on actual project state:

{
  "situation": "3 jobs failed in last 24h",
  "recommendations": [
    {"priority": 1, "action": "kbagent job detail --project prod --job-id 789", "reason": "Most recent failure, same component as other 2"},
    {"priority": 2, "action": "kbagent config detail ...", "reason": "Config was modified 2h before failures started"}
  ],
  "pattern": "2/3 failures are in mysql-extract component — likely a connection issue"
}

Open questions

• Should this use the AI Service API for reasoning, or be rule-based?
• Scope: project-level only, or cross-project?
• How opinionated should suggestions be?

Context

Discussion from Devil's Advocate analysis of kbagent's agentic capabilities.

[padak]

Part of #152 (Modern Agentic CLI epic).