[cli] better logging

[spencatro-pub]

I keep needing to add silly hacks to my hooks to better understand what's going on when an agentblame call fails, for example:

{
  "version": 1,
  "hooks": {
    "preToolUse": [
      {
         "bash": "AGENTBLAME_DEBUG=1 agentblame ... >> /tmp/agentblame.log 2>&1",
      }
    ]
}

Does anyone have a better strategy for reviewing logs when something goes wrong? Maybe I just need to update my process, but if not, now that v3 has a config system, maybe a config item to write agentblame logs to disk instead of being lost to stdout would be useful? (This probably interacts with --trace-hooks and --strict, but as an outside user, I should probably explain needs rather than dictate design.)

[muraalee]

Will look into this.