docs: capabilities matrix — add the Engine path column

[Kiran01bm]

Adds an "Engine path" column to every table in docs/capabilities.md, stating how the engine carries (or will carry) each operation, alongside the existing Status column that says what happens today.

Why

Status answers "what happens if I submit this today" but not the question users ask next: "how will pg-sprite actually run it?" Two operations with the same 🟡 status can have very different futures — one becomes a native safer sequence, another needs the full copy-and-swap machinery. Making the path explicit also gives each planned refusal a stated destination, so the matrix doubles as a roadmap.

What

• New "The engine path" subsection defining the five path values: native, as-is · native, safer sequence · native, planned flow · copy-and-swap · — (no path), with links to safer-sequences.md and optimistic-attempt.md.
• Every row in all seven tables gains an Engine path value (10 as-is / 7 safer sequence / 11 planned flow / 8 copy-and-swap / 15 no path = 51).
• Headline counts corrected: 19 planned behind a typed refusal and 13 out of scope by design (previously stated 18/14, which did not match the tables).

Before / after

before:
  | Operation | Status | Why |
  | DROP COLUMN | 🟡 planned | needs a safer sequence |
        └── reader knows it refuses today; not how it will eventually run

after:
  | Operation | Status | Engine path | Why |
  | DROP COLUMN | 🟡 planned | native, safer sequence | needs a safer sequence |
        └── status = today's behavior; path = the mechanism, present or planned

---

This PR was authored by Amp (Claude Opus 4.5) on Kiran's (@Kiran01bm) behalf.

[chatgpt-codex-connector]

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.