Define the v0.0.3 user tutorial ladder

[adohe]

Spec candidate: D1 (AReno v0.0.3 local iteration spec).

Scope

• Audit current README, Sphinx docs, and examples.
• Define the tutorial ladder users should see for local post-training: install/check, tiny smoke, dataset/reward, SFT/DPO, GSPO/GRPO, agentic rollout, serving, and observability.
• Identify which tutorial should be written first in v0.0.3.

Acceptance

• The plan references current docs and examples.
• Each proposed tutorial maps to one small follow-up issue.
• The plan does not promise full TRL/Tinker parity in v0.0.3.