[Feature]: clean up stale github-copilot gpt-5-nano references

[divlook]

Problem Description

The current opencode models github-copilot --verbose output does not expose a github-copilot/gpt-5-nano model, but the repository still contains stale references to github-copilot/gpt-5-nano in fallback expectations and snapshots.

This creates confusion when reviewing or debugging GitHub Copilot model selection because snapshot-based or fallback-based expectations do not match the current catalog output.

Proposed Solution

Clean up stale github-copilot/gpt-5-nano references and align GitHub Copilot-specific expectations with the current opencode models github-copilot --verbose output.

Suggested scope:

• update stale GitHub Copilot nano references in snapshots and tests
• review fallback expectations that still assume github-copilot/gpt-5-nano
• update any user-facing docs or examples if they imply GitHub Copilot still exposes that model
• keep provider-specific cleanup scoped so unrelated openai/gpt-5-nano or opencode/gpt-5-nano references are only changed if they are also stale

Feature Type

Other

Alternatives Considered

• leave the stale references in place and treat them as historical expectations
• fix them opportunistically in unrelated PRs

Both options make future debugging harder and increase the chance of catalog drift going unnoticed.

Additional Context

Evidence from current catalog output:

• opencode models github-copilot --verbose shows GPT-5 entries such as gpt-5-mini, gpt-5.1, gpt-5.2, gpt-5.2-codex, gpt-5.3-codex, gpt-5.4, and gpt-5.4-mini
• no github-copilot/gpt-5-nano entry appears in that output

Examples of stale references observed during the Copilot variant alignment work:

• src/cli/__snapshots__/model-fallback.test.ts.snap
• related fallback expectations or docs that still mention github-copilot/gpt-5-nano

Contribution

• I can help submit a PR for this

[divlook]

Opened as a follow-up to PR #3353. That PR fixes the GitHub Copilot variant alignment work; stale references were intentionally split into this separate cleanup issue to keep the PR scope focused.

[divlook]

Clarification: this follow-up specifically tracks stale github-copilot/gpt-5-nano references that no longer match the current opencode models github-copilot --verbose output.