[ML] Transforms: add throttling UI support

[hendrikmuhs]

depends on: elastic/elasticsearch#54862

The backend will add support for optional throttling, this is a numeric setting: requests_per_second.

Discuss: A slider would be nice, however it's unclear what the max is. I could add a recommendation in the _preview API by calculating an average requests per second from existing transforms.

[elasticmachine]

Pinging @elastic/ml-ui (:ml)

[walterra]

The feature to edit throttling is now in the UI (#65365).

@hendrikmuhs Would be great if you get a chance to test it, please also have a look if you're happy with the wording in the form. At the moment it's not very obvious to users which values might be good for tuning the throttling.

I could add a recommendation in the _preview API by calculating an average requests per second from existing transforms.

Is that something you got around to do or still plan to do?

[hendrikmuhs]

I could add a recommendation in the _preview API by calculating an average requests per second from existing transforms.

Is that something you got around to do or still plan to do?

I still think this is a good idea and we should at least prototype it, for a future version, not 7.8.

[hendrikmuhs]

Tested it, looks great.

Note that update allows more:

• changing source: useful regarding the query, e.g. to filter out noise
• changing dest: useful when the user realizes that he wants to use ILM and set an alias here

So this is another reason for a json editor component or adding the specific parts, which might get complicated. As a mitigation until that is in place, we could add a sentence on the top, where you already link the docs.

[walterra]

Thanks for the feedback! I included it in the follow-up meta issue about editing transforms here: #67255