Fix race condition in EventManager refresh error handling

[Copilot]

When multiple refreshes fired concurrently, a failing refresh could overwrite newer values written by a succeeding one — because each Task captured a current snapshot upfront, and the error handler blindly restored that stale state.

Changes

• EventManager.swift: On fetch failure, emit nil instead of the stale current snapshot; add .compactMap { $0 } so the .sink only fires on success. Failed refreshes now leave published properties untouched entirely.

// Before: stale snapshot could overwrite newer successful state
let current = await MainActor.run { (self.calendars, self.events) }
// ...
} catch {
    promise(.success(current)) // race: may clobber a later successful refresh
}

// After: failed refresh is a no-op on published state
} catch {
    promise(.success(nil))
}
// ...
.compactMap { $0 }  // skip nils — properties unchanged on error

• FakeEventStore.swift: Added shouldThrow flag to simulate fetch failures in tests.
• EventManagerTests.swift: Added test_eventsPreservedOnRefreshError — uses an inverted expectation to assert no $events update is published after a failed refresh.

---

💡 You can make Copilot smarter by setting up custom instructions, customizing its development environment and configuring Model Context Protocol (MCP) servers. Learn more Copilot coding agent tips in the docs.