/datasets/list result format #2599
Description
The /datasets/list API returns {"name": dataset.name, "controller": dataset.controller, "run_id": dataset.md5} even though we've "demoted" the concept of controller.
This is partly because of the metadata mapping of Dataset SQL object columns, which currently does not include either column. (See #2597). Once we have those available, it would be cleaner to drop "controller" (at least) from the default response and allow the caller to ask for the metadata if necessary.
We also need to migrate from using "run_id" in the "second-level" APIs (TOC, sample, timeseries data), in which case that information will be of less general use and arguably should also be demoted to optional metadata.
This would effectively reduce the default output to a list of dataset names that match the search critera specified; any additional metadata requested would be through the "metadata" keyword.