Jay Narale

You mean push to this PR? Its a fork branch and I don't think we can

@steveburnett ``` Iceberg Changes * Add Manifest file caching support for Iceberg Native Catalogs :pr:`21399` * Fix Iceberg memory leak with Delete File. :pr:`21612` General Changes * Add function map_top_n_keys_by_value...

I believe we should introduce a customizable constant, which can be used to scale the projected upper limit, particularly when statistics are not accessible.

@fgwang7w do you need any help in merging/testing this PR?

Overall looks good, @tdcmeehan can you please review the SPI changes. This PR doesn't include loading the histograms from connectors and the SPI changes are currently noop. So, I'm uncertain...

@Yuhta Is there any documentation on what **LazyVector** exactly is? Wanted to understand how it is decreasing the data read

I saw the comment [here](https://github.com/facebookincubator/velox/blob/17c3d8480ffee051252dcb7b61ce1fafcb0ec6ba/velox/vector/LazyVector.h#L223). From the comments looks like if there is filter pushdown in scan, lazy loading can decrease some loading into memory. However I am confused how...

got it thanks!