FR: Run mode enhancements

[predictiple]

1. When the --api_config flag is used with -r and the api config is either invalid or not found then the artifacts collect runs locally without any warnings or errors. So we probably need some stronger validation/signalling around that.

2. When --client is used and the client is offline, then the command waits with no feedback to the user. By design it's meant to run synchronously, but maybe it would be better if an offline client caused it to just return the flow ID and exit.

3. If the client is online but the collection is long-running then it would be good to have some sort of feedback in the terminal. Maybe show the state of the flow, if that's possible? Otherwise there's no way to tell the difference between an offline client situation and one where the collection is just taking long.

[scudette]

I dont know if changing the behavior depending on if the client is online is a good idea - I would imaging the command line option being used in e.g. a bash script and so should behave consistently. Maybe a behavior change could be implemented with an additional flag but making it asynchronous with a flag is about the same as the collect() API call anyway.

As for point 3 - maybe we should also stream any query logs over the API so it is clear that the client has started the collection?

[predictiple]

Streaming the query logs sounds helpful, but there are no logs until the collection starts, so the progress is in the dark until that happens. I was imagining polling and returning the collection state, as shown in the GUI.

An async flag would be useful. But in both sync or async situations it would be great if it emitted a flow ID as early as possible - that way if the user decides to Ctrl+C, or perhaps it's a script that has hit it's own timeout, then it doesn't require follow-up queries to figure out what flow was created. The follow-up process can then just go straight to artifacts fetch.