Allow concat_batches to take non owned RecordBatch

[alamb]

Is your feature request related to a problem or challenge? Please describe what you are trying to do.

We have code that has Vec<&RecordBatch> (not Vec<RecordBatch>) and thus we can't call concat_batches as that requires taking ownership. I looked at the code for concat_batches and it is not at all clear why it needs ownership of the RecordBatch

https://docs.rs/arrow/30.0.0/arrow/compute/fn.concat_batches.html

Describe the solution you'd like
I would like some way to call concat_batches that doesn't require an bunch of owned RecordBatches

Perhaps something like this would work:

pub fn concat_batches(
    schema: &Arc<Schema>,
    batches: impl IntoIter<Item = &RecordBatch>
) -> Result<RecordBatch, ArrowError>

And existing callers could call concat_batches(&batches)

Describe alternatives you've considered

Additional context
in a DataFusion PR: apache/datafusion#4777 (comment)

[askoa]

I can pick this up. I briefly looked into this. concat_batches uses iter() in multiple places in the function. It also passes the iter() to another function. into_iter() is consuming and hence cannot be used multiple times within the function.

I am thinking of creating a Vec using the into_iter() as the first step in the function and then use that Vec's iter() in the rest of the function. Let me know if you have different suggestion.

Edit: One more question is concat_batches part of public api?

[alamb]

I think collecting references into a Vec<&RecordBatch> seems like a very reasonable step to me

[tustvold]

label_issue.py automatically added labels {'arrow'} from #3480