Performance: MockBot restarts Actix server on every dispatch() call

[stepan-romankov]

Hi! First of all, thanks for creating this testing library - it's been very helpful for testing teloxide bots.

I noticed that MockBot::dispatch() creates a new ServerManager instance on every call, which starts and stops an Actix HTTP server for each dispatched update:

pub async fn dispatch(&mut self) {
    self.state.lock().unwrap().reset();

    let server = ServerManager::start(self.me.clone(), self.state.clone())
        .await
        .unwrap();  // New server every time

    // ... dispatch logic ...

    server.stop().await.unwrap();  // Server stopped
}

For integration tests that simulate realistic user flows (e.g., multi-step registration, ordering workflows), this adds significant overhead. A test with 50+ dispatch calls can take several minutes due to repeated server startup/shutdown.

Questions:

Is there a specific reason the server is restarted on each dispatch? (e.g., state isolation, thread safety concerns)
Would you be open to a PR that keeps the server alive across dispatches and only resets the response state?
Proposed change:

The server could be started lazily on first dispatch and reused for subsequent calls:

pub async fn dispatch(&mut self) {
    // Only reset responses, not the entire server
    self.state.lock().unwrap().responses = Responses::default();

    // Start server only if not already running
    if self.server.is_none() {
        self.server = Some(ServerManager::start(...).await?);
    }

    // ... dispatch using existing server ...
    
    // Server stays alive, stopped on MockBot::drop()
}

This would preserve the existing API while significantly improving performance for multi-dispatch test scenarios.

I've prototyped this approach locally and saw test times drop from ~60s to ~1s for a 20-dispatch test. Happy to submit a PR if you think this direction makes sense.