fix(registry): catch up to @w3s/playground-registry v11

[ottovlotto]

Summary

Tonight's nightly (#207) failed across five cells on Revive.ContractReverted from registry.publish(). Root cause: the contract was redeployed in playground-app PR #204 ("feat(registry): points, leaderboard, star toggle, mod credit") with three new required tail params. The CLI's pinned ABI was still v7's four-arg signature, so every publish call reverted on signature mismatch.

This PR catches the CLI up.

Changes

cdm.json

Replaces the @w3s/playground-registry entry under target 929b5c63e2cb5202 with the v11 block ported verbatim from playground-app's cdm.json (same target hash, same Paseo Next v2 chain endpoints — only contract address, version, metadataCid, and ABI change).

• Before: 0xea3fb6C5cDA79FEEf5b2d42a9122fC1BAFaF647F v7
• After: 0x109BBf68F41B570Fd6d3b7bDE9b8e18779FDd8Db v11

src/utils/deploy/playground.ts

registry.publish.tx(...) now passes the three new tail params:

New param	Value	Why
modded_from	""	Capture-at-mod-time isn't yet wired up in the CLI (V1 P0 per spec). Empty string is the contract's sentinel for "no source".
is_moddable	repositoryUrl !== null	Same signal the app metadata already carries — true for --moddable deploys.
is_dev_signer	publishSigner.source === \"dev\"	True for Alice / --suri flows. The contract uses this to skip the launch-points award when a shared dev key signs on the user's behalf.

src/utils/deploy/playground.test.ts

Three existing toHaveBeenCalledWith assertions now check the full 7-arg signature. Adds one new test ("passes is_dev_signer=true and the right is_moddable when the signer source is dev") that exercises both new param variants in a single publish, paired with is_moddable=false, so a future regression hard-coding either to a default would surface immediately.

Follow-up not addressed here

• modded_from capture-at-mod-time. `dot mod ` needs to write the source domain to local metadata, threaded to publishToPlayground and passed here. V1 P0 per the spec — separate PR.

Test plan

• pnpm format:check — clean
• pnpm lint:license — clean
• pnpm build — clean
• pnpm test — 497 passed / 1 skipped
• workflow_dispatch nightly on this branch — moddable cell goes green on first attempt
• Subsequent scheduled nightly green; Nightly E2E failure: 2026-05-26 #207 auto-closes

Closes #207

[UtkarshBhardwaj007]

@ottovlotto already merged this in my PR

[ottovlotto]

@UtkarshBhardwaj007 @charlesHetterich — PM-initiated starter PR to unblock tonight's nightly (#207). Bundles three things tightly because they have to ship together: cdm.json refresh to v11 (ported from playground-app), the new publish() tail args in src/utils/deploy/playground.ts, and matching test updates.

The src/utils/deploy/playground.ts change is CLI team territory — is_dev_signer, is_moddable, and modded_from semantics. I made the most direct mapping I could from what was already in scope (publishSigner.source === "dev", repositoryUrl !== null, and "" for not-yet-built modded-from capture), but you're better positioned to judge whether that's right — particularly is_dev_signer since you touched the dev-publish path most recently in #193.

Action options, all good from my end:

1. Merge as-is if the approach checks out. Closes Nightly E2E failure: 2026-05-26 #207, nightly comes back tomorrow.
2. Modify in place — push directly to this branch if it needs tweaks.
3. Take your own approach in a separate PR and close this. If you do, I'll come back and update the test-side coverage in a follow-up PR once your CLI change lands and we know what shape is_dev_signer / is_moddable / modded_from actually want to be.

No urgency from my side on which option — happy to leave it parked until you have a moment. CI is currently degraded (workflow_dispatch returning 500 since ~08:00 UTC; PR-event workflows aren't firing either) so live verification of the publish path against the new contract is on hold regardless.