mason v0.6.1

Patch Changes

• #19 591ce4c Thanks @antstanley! - fix: the wasm service worker survives a deploy. Each deploy deletes the previous
  build's hashed assets, so a worker that installed before a deploy would 404
  fetching its old wasm engine and then brick every /api/feed for the life of
  that session (the rejected init was memoised) — a 500 that only hit visitors who
  had loaded the app before. The worker now precaches its own wasm and loads the
  engine from that cache, so it keeps serving until it is itself replaced; a failed
  init is never memoised. The client also revalidates the worker script on load
  (updateViaCache: 'none' + update()) and reloads once when a new engine takes
  control, so a deploy is picked up promptly instead of leaving a stale worker.