I made two additional changes to Dispatcher.update_persistence:

1. Use a lock to ensure that the method is not called twice simultanuously
2. convert dispatcher.user/chat_data.keys() to lists to prevent errors like this. That particular one was probably caused by JobQueue calling Dispatcher.update_persistence while an entry was added to dispatcher.user_data, but similar situations can probably also arise when calling Dispatcher.update_persistence from the async threads. We don't acutally have to worry about the new entry not being persisted, as update_persistence() will be called for the update that lead to the new entry anyway.

Could this async be added to job callback as well? I'm using @run_async decorator to run jobs in parallel for now

@GauthamramRavichandran Please have a look at this comment :)

As discussed, I restructured a bit. Let me try to summarize:

• The error_handler parameter for run_async was basically there to avoid infinite loops coming from asynchronously run functions error handlers
• instead, now there is just a parameter async_error_handling. When False, no async error_handler will be called*
• but we don't even handle sync errors raised by error handlers. So now there is just a parameter error_handling for run_async. When False, the error will just be logged.
• We only have one set of error handlers, namely those registered with add_error_handler
• new parameter async_params for CallbackContext so that the params for custom async functions can be accessed from the error handler