Make the first README demo prove tests and typecheck can pass while CellFence fails #19

Problem

The first README flow is currently a successful cellfence init / cellfence check path. That is useful, but it does not immediately demonstrate CellFence's core value: ordinary project checks can pass while an architectural boundary violation is still blocked.

A first-time reader should be able to see, near the top of the README, that CellFence catches a private cross-cell import even when the code still typechecks and tests pass.

Expected behavior

Update the first runnable README demo so it shows this sequence:

1. project tests pass
2. TypeScript typecheck passes
3. CellFence fails with CELLFENCE_PRIVATE_IMPORT

The demo should make the contrast obvious without requiring a large fixture or prior CellFence knowledge.

Example intent:

npm test
# passes

npm run typecheck
# passes

npx cellfence check
# fails with CELLFENCE_PRIVATE_IMPORT

The exact commands can differ if the demo remains copy-pasteable and accurate.

Scope

Keep this focused on README onboarding.

Do not change CellFence rule behavior.
Do not change the manifest schema.
Do not add a large new example application unless it is clearly needed for the README flow.

Suggested starting points

Likely areas:

README.md
examples/minimal/README.md
examples/minimal/cellfence.manifest.json

The existing README sections to review are:

Try it in sixty seconds
Catch a violation in thirty seconds

Acceptance criteria

• The first README demo clearly shows tests passing, typecheck passing, and CellFence failing.
• The failure shown is CELLFENCE_PRIVATE_IMPORT.
• The example includes enough files or commands for a reader to reproduce it.
• The README still keeps a short successful init/check path, either before or after the failure demo.
• No product behavior is changed.
• No generated benchmark or research output is added.

Validation

Run:

npm run build
npm test

If the README includes a copy-paste shell demo, manually run that demo in a temporary directory and report the result in the PR.