v1.0.4: tighten offline tests and adoption diagnostics

[jeffhuber]

Goal

Resolve the three concrete cleanup findings from the v1.0.3 multi-provider adoption round without changing runtime policy or the privacy boundary.

Scope

• Make the standalone-wrapper reinstall fixture deterministic without package-index access. The default contributor test path must remain offline-friendly.
• Add explicit configuration-source classification to setup-drift text output, matching the existing structured report and init terminology.
• Surface the adoption posture hint early enough that hosted/orchestrator operators do not mistake default local-provider warnings for their intended posture.
• Add one concise documentation note where behavior changes.

Acceptance criteria

• The relevant wrapper test passes with package-index access disabled.
• Normal unit tests do not require PyPI, TestPyPI, or a live provider.
• Setup-drift text names packaged starter versus explicit repository configuration.
• Adoption doctor output presents the posture hint before provider-specific warning detail when it is applicable; JSON IDs and ordering remain stable and tested.
• Focused tests, full tests, static checks, privacy scan, and smoke checks pass.
• No generated workflow, gate, token path, event schema, or upload behavior changes.

Parent: #655

Code Mower execution

Builder experiment arm: Muse CLI. Use an isolated worktree, record elapsed time and interventions, and send the resulting PR through both Codex and Claude peer audits. Fix or explicitly decide every BLOCKED finding.