Activation task deduplication and SharedActivationCoordinator enhancements

[CharlVS]

Copied from upstream PR GLEECBTC#274

closes GLEECBTC#273

Summary

• Ensures a single activation per AssetId across the SDK by routing all public activation calls through SharedActivationCoordinator and making ActivationManager idempotent at init.
• Prevents balance/pubkeys from triggering overlapping coin-init while activation is pending by joining the in-flight activation.
• Hardens event streaming to avoid WASM closure re-entrancy and post-dispose callbacks.

Acceptance criteria

• One activation task per asset across concurrent callers.
• No IndexedDB/storage uniqueness violations during activation + balance races.
• No re-entrancy/after-dispose WASM closure errors under stress.
• Activation responses parse reliably across protocols.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Close activation stream controller so callers can complete

When SharedActivationCoordinator is available, activateAssets now creates a broadcast controller and forwards each asset’s activation stream, but the controller is never closed or completed once the forwarded streams finish. As a result the returned stream never emits done, so any caller awaiting completion (e.g. await activateAssets([...]).last) will hang indefinitely and the controller/listeners stay live even after all activations end. The old implementation from ActivationManager.activateAssets completed normally, so this is a regression.

Useful? React with 👍 / 👎.