feat: discover Muse Wonder and Wonder Full subscription models

[kato-mg]

Summary

• Admit documented muse-wonder-1.0 and muse-wonder-full-1.0 IDs only when returned by account-visible discovery.
• Preserve Spark/Contributor discovery and fallback; exclude non-chat families and Wonder lookalikes.
• Document Full entitlement, account-level Contributor settings, and model selection versus reasoning effort.

Verification

• 15 offline tests passed, independently rerun on commit 9a0567b.
• Plugin validator and git diff --check passed.
• No authentication, transport, endpoint, dependency or reasoning changes.
• No live catalogue/inference or account-entitlement verification; bounded discovery fix.

[TheStreamCode]

Thank you for this: the exact-ID approach, the lookalike tests, and the subscription and entitlement docs are solid work, and the Wonder IDs plus 1M context check out against Meta docs. I am closing in favor of a different direction: model discovery is moving to a fully dynamic list (whatever the live account catalog returns, minus known non-chat families), so new models appear without per-model PRs. Your verification notes helped confirm the Wonder data. Please test the dynamic release when it lands.

[TheStreamCode]

v0.3.0 is out with fully dynamic discovery: every id from the live catalog is served except the verified non-chat families (sam-, muse-image-, muse-voice-*), so the Wonder IDs from your verification notes resolve with no per-model entries. Your stated caveat — no live catalogue verification — is exactly where help would be most valuable now: if you can test v0.3.0 against a live subscription, that confirmation would be appreciated. Separately, if any of your entitlement/Contributor-settings notes didn't make it into the rewritten README, a focused docs PR would be welcome — happy to review it. Thanks again for the thorough verification work.