This is fix to issue described in: [ENG-5906](https://linear.app/comm/issue/ENG-5906/web-app-stuck-in-crash-loop-even-after-refreshing#comment-11a5f277).
In D9948 I started to persist `updatesCurrentAsOf`.
I also updated `setInitialReduxState` action to rely on rehydrated data when it's a staff user.
What I missed is updating `currentAsOfPromise`, which for `updatesCurrentAsOf` was using server time [here](https://github.com/CommE2E/comm/blob/f4e61ed90d7f972909d2372a397ee3da5fdbd17e/keyserver/src/responders/redux-state-responders.js#L306).
This means that after reloading the app, the client (staff) has its own `updatesCurrentAsOf` from the previously received update (previous session) while the keyserver has newer `updatesCurrentAsOf` created while returning the initial state version.
This causes [this condition](https://github.com/CommE2E/comm/blob/f4e61ed90d7f972909d2372a397ee3da5fdbd17e/keyserver/src/socket/session-utils.js#L340C10-L340C10) to be `true`, and as a result was causing `FULL_STATE_SYNC` [here](https://github.com/CommE2E/comm/blob/f4e61ed90d7f972909d2372a397ee3da5fdbd17e/keyserver/src/socket/socket.js#L510C19-L510C19) which on prod kills the socket because of the amount of data to transfer.
This diff should make `updatesCurrentAsOf` synced between client and server.
This also fixes issue described [here](https://linear.app/comm/issue/ENG-5906/web-app-stuck-in-crash-loop-even-after-refreshing#comment-92c744ce). Previously, when `updatesCurrentAsOf` was just started to be persisted, it gets rehydrated with `undefined/0` value, this means after loading app all updates were started to be fetched. Right now if `updatesCurrentAsOf` is `0` we fetch everything in `InitialReduxStateResponder` and set `updatesCurrentAsOf` as `serverTime`.
This is only for staff, for standard users everything should work as previously.