refactor: centralize DEFAULT_MODEL constant, fix remaining kimi-k2.5-free references (#208)

[konard]

Summary

• Root cause (primary): hive-mind's agent.lib.mjs passes agentArgs as a single template-interpolated string (e.g., --model opencode/minimax-m2.5-free --verbose) to command-stream's $ tag. The command-stream quote() function wraps this space-containing string in single quotes, making the agent receive it as one argument — so yargs and getModelFromProcessArgv() never see --model as a separate flag, and the agent falls back to the yargs default.
• Root cause (secondary): The yargs default in index.js was still opencode/kimi-k2.5-free, so the fallback was wrong.
• Fix (Phase 2): Updated yargs default in index.js from kimi-k2.5-free to minimax-m2.5-free.
• Fix (Phase 3): Centralized the default model as DEFAULT_MODEL constant in js/src/cli/defaults.ts — all places that reference the default now import from this single source.

Changes

• js/src/cli/defaults.ts — New file: exports DEFAULT_MODEL = 'opencode/minimax-m2.5-free', DEFAULT_PROVIDER_ID, and DEFAULT_MODEL_ID constants as the single source of truth
• js/src/index.js — Updated --model yargs default from opencode/kimi-k2.5-free to DEFAULT_MODEL (via import from defaults.ts)
• js/src/cli/model-config.js — Updated --use-existing-claude-oauth default-model check and error messages to use DEFAULT_PROVIDER_ID/DEFAULT_MODEL_ID (via import from defaults.ts); previously referenced kimi-k2.5-free
• js/src/tool/task.ts — Updated fallback model for sub-agent tasks to use DEFAULT_PROVIDER_ID/DEFAULT_MODEL_ID (via import from defaults.ts); previously referenced minimax-m2.5-free as a literal
• js/tests/models-cache.test.js — Fixed echo provider test to explicitly pass --model link-assistant/echo
• docs/case-studies/issue-208/ — Added second failing run analysis with correct root cause (command-stream quoting bug), Phase 2/3 documentation, and the second log file

Test plan

• npm run check (eslint + prettier) passes
• bun test tests/models-cache.test.js — 5 pass, 0 fail
• bun test tests/model-fallback.test.ts tests/model-not-supported.test.ts tests/output-response-model.test.js tests/model-validation.test.ts tests/models-cache.test.js — 41 pass, 0 fail

Fixes #208

🤖 Generated with Claude Code

[konard]

🤖 Solution Draft Log

This log file contains the complete execution trace of the AI solution draft process.

💰 Cost estimation:

• Public pricing estimate: $2.898786
• Calculated by Anthropic: $4.761121 USD
• Difference: $1.862335 (+64.25%)
  📎 Log file uploaded as Gist (5095KB)
  🔗 View complete solution draft log

---

Now working session is ended, feel free to review and add any feedback on the solution draft.

[konard]

🔄 Auto-restart 1/3

Detected uncommitted changes from previous run. Starting new session to review and commit them.

Uncommitted files:

?? gist-log.txt

---

Auto-restart will stop after changes are committed or after 2 more iterations. Please wait until working session will end and give your feedback.

[konard]

🔄 Auto-restart 1/3 Log

This log file contains the complete execution trace of the AI solution draft process.

💰 Cost estimation:

• Public pricing estimate: $6.208658
• Calculated by Anthropic: $4.575421 USD
• Difference: $-1.633237 (-26.31%)
  📎 Log file uploaded as Gist (8541KB)
  🔗 View complete solution draft log

---

Now working session is ended, feel free to review and add any feedback on the solution draft.

[konard]

✅ Ready to merge

This pull request is now ready to be merged:

• All CI checks have passed
• No merge conflicts
• No pending changes

---

Monitored by hive-mind with --auto-restart-until-mergeable flag