Nok Lam Chan

@lucasjamar Hi Lucas, just checking in to see if you need any help with this.

@lucasjamar No worries! Looking forward to see your PR😀.

@lucasjamar Are you still interested in this?

@lucasjamar Hey thanks for making time for this! Do you want to make this a PR? It would be easier for me to drop comments on it. :)

At a glance making just NetCDF works is straight forward, but seems @lucasjamar is going with a GenericXarrayDataSet. I wonder if it makes sense to take the `pandas` approach with...

I am actually not aware of the pandas.GenericDataSet😅 do people actually use that and do we keep both for backward compatiblitity reason? I think it is possible since the API...

Not sure if I like the idea of deprecating the dedicated one, but this is not urgent so I will think more about it. I added the `dataset` related refactoring...

Adding this while I am looking at #1768 some object store related issues. Currently, we actually put this into `CustomDataSet` where it needs to access the `self._fs`, at least 99%...

Added [Expose version load version information when load_version=None](https://github.com/kedro-org/kedro/pull/1911) Another symptom here is: * `self.resolve_load_version` isn't doing anything, the real function that we want people to use is `self._get_load_path`, but it's...

1. Generally agree, I don't think the implementation will be complicated, we just need to decide how we expose this feature to our user. The downside of using `hook` is...