Define Base test-first and review-feedback guidance

[codeforester]

Problem

The obra/superpowers TDD and code-review skills are stricter than Base should adopt wholesale, but the underlying practices are valuable for Base behavior changes: write regression tests for bugs, prove tests fail for the right reason, and evaluate review feedback against actual Base constraints before implementing it.

Reference:

• https://github.com/obra/superpowers/blob/main/skills/test-driven-development/SKILL.md
• https://github.com/obra/superpowers/blob/main/skills/requesting-code-review/SKILL.md
• https://github.com/obra/superpowers/blob/main/skills/receiving-code-review/SKILL.md

Desired Outcome

Document a Base-specific test-first and review-feedback workflow for user-facing command behavior, shared runtime behavior, and bug fixes.

Scope

• State when Base changes should start with a failing test or reproduction.
• Explain how to prove a regression test catches the bug before relying on it.
• Clarify exceptions for docs-only changes, prototypes, or changes where automated tests are not practical.
• Add guidance for evaluating external review feedback before implementing it.
• Tie examples to Base's BATS, Python pytest, integration tests, and basectl test base conventions.

Non-Goals

• Do not require strict TDD for every documentation-only or configuration-only change.
• Do not import Superpowers tone rules or human-interaction rules.
• Do not broaden test requirements beyond the risk of the change.

Acceptance Criteria

• Base guidance explains test-first expectations for behavior changes and bug fixes.
• PR validation guidance distinguishes regression proof, focused checks, and broader suites.
• Review-feedback guidance says to verify the suggestion against Base's architecture and product boundaries before implementing it.
• The guidance keeps test coverage proportional to risk and blast radius.

Validation

• Run git diff --check.
• If test workflow docs are changed substantially, run any existing docs or workflow validation that covers them.

Demo Impact

None unless the change affects demo validation guidance.