fix(harness): match Bedrock GPT-6 effort levels to what each model accepts

[Katsuhito04]

Human Overview

Lets GPT-6 Sol, Luna and 6.1 Sol on Bedrock take an effort level, and makes effort "off" work on Astra and 6.1 Sol, which reject "none".

Description

The Bedrock GPT effort table only listed GPT-6 Astra, and it gave Astra a none level that Bedrock rejects. Sol and Luna now use the existing GPT levels; Astra and 6.1 Sol use a set without none, so off omits the reasoning field like GPT-OSS. The CLI's own effort menu will follow once a harness release includes this.

Opened as a draft so the level split can be confirmed before review; it changes what off sends for Astra.

Resolves #5174

This PR was prepared with help from an AI coding assistant; I reviewed and understand every change.

Related Issues

#5174

Documentation PR

None needed.

Type of Change

Bug fix

Testing

Ran npm run check in harness-ts and ruff plus pytest in harness-py. Against Bedrock (us.openai.*), resolved each model through harness-py and ran an agent turn: Astra off, 6.1 Sol off and max, Sol high, Luna off all succeed (on main, Astra off fails with ValidationException and Sol high / 6.1 Sol max are rejected).

• I ran hatch run prepare

Checklist

• I have read the CONTRIBUTING document
• I have reviewed and understand every line of code in this PR, including any generated by AI tools, and I can explain why it works
• My change is focused and reasonably small; I have split unrelated work into separate PRs
• I have added any necessary tests that prove my fix is effective or my feature works
• I have updated the documentation accordingly
• I have added an appropriate example to the documentation to outline the feature, or no new docs are needed
• My changes generate no new warnings
• Any dependent changes have been merged and published

---

By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.