load_cdf return an empty dataframe when a version is out of range

[pblocz]

Description

In spark delta table you can enable an option to manage out of range versions or timestamps. https://docs.delta.io/latest/delta-change-data-feed.html#read-changes-in-streaming-queries

Right now the behaviour of load_cdf is inconsistent, if you provide an out of range version you get an error:

But with a timestamp out of range, you get an empty dataset:

It would be useful for incremental pipelines to have a way to manage this behaviour and make it consistent.

[ion-elgreco]

If you know some rust it's probably a simple fix

[pblocz]

If you know some rust it's probably a simple fix

@ion-elgreco I have never used rust and it is been a while since I have done anything that needs to be compiles, but can give it a go