## Current Behavior

When the daemon dies while processing a request, the reconnect logic
adds the retry back to the promise-based queue. However, the original
request is still blocked in the queue waiting for a response that will
never come (the socket is dead). This creates a deadlock:

1. Original request (`fn1`) is blocked awaiting a promise that will
never resolve
2. Retry request (`fn2`) is queued but can't execute until `fn1`
completes
3. `fn1` can't complete because it's waiting on the dead socket

## Expected Behavior

When reconnecting after daemon death, the retry should resolve the
pending promise that the original queue entry is waiting on, allowing
the queue to proceed normally.

## Related Issue(s)

<!-- No specific issue, discovered during development -->

## Solution

Instead of re-queuing the retry through `sendToDaemonViaQueue` (which
adds to the end of the queue), we now call `sendMessageToDaemon`
directly. This resolves the pending promise that `fn1` is waiting on,
allowing it to complete naturally and the queue to proceed.

Also removed the now-unused `decrementQueueCounter` method from
`PromisedBasedQueue`.

(cherry picked from commit 89aa25e)