Utilized column analyzer unnest fix map case

[kevintang2022]

Description

Currently, unnest cannot handle input expresssions that are the Map type. This is because 2 output expressions are created when unnest is used on a Map (one for key and one for value). This leads to index out of range errors because the current code only expects that Array types are passed into unnest.

Here is an example of building a mapping from output column to its corresponding input expression in the unnest.

unnest(m1, a1, m2, a2, m3,m4) t (m1k, m1v, a1v, m2k,m2v, a2v, m3k, m3v, m4k, m4v)
The correct field to expression index mapping should be
{
m1k: 0
m1v: 0
a1v: 1
m2k: 2
m2v: 2
a2v: 3
...
}

but it is currently

{
m1k: 0
m1v: 1
a1v: 2
m2k: 3
m2v: 4
...
}

Motivation and Context

T217212061
Currently, for utilized column analyzer, unnest will throw an index out of range exception when a map column is passed into the unnest function. This is important because if an index error is thrown, all columns in utilized tables will be checked for ACL.

Impact

With this change, utilized column analyzer will check all expressions passed to unnest instead of checking all columns in the entire query. This will unblock cases like unnest(m) t (c1, c2)

Test Plan

Unit tests in utilized column analyzer covers the cases mentioned, which were failing before.

Contributor checklist

• Please make sure your submission complies with our contributing guide, in particular code style and commit standards.
• PR description addresses the issue accurately and concisely. If the change is non-trivial, a GitHub Issue is referenced.
• Documented new properties (with its default value), SQL syntax, functions, or other functionality.
• If release notes are required, they follow the release notes guidelines.
• Adequate tests were added if applicable.
• CI passed.

Release Notes

Please follow release notes guidelines and fill in the release notes below.

== RELEASE NOTES ==

General Changes
* Fix index error when a map column is passed into an unnest function by using the column analyzer to correctly map key and value output fields back to correct input expression

[kevintang2022]

This is the type of cases that will experience some regression, but it should easy for the user to unblock themselves if they can just clean out the unused expressions passed into unnest

[steveburnett]

Thanks for the release note entry! Suggest rephrasing to follow the Order of Changes in the Release Notes Guidelines. I adapted phrasing from your Description in the PR, but if you have a better way to phrase it, let me know.

== RELEASE NOTES ==

General Changes
* Fix index error when a map column is passed into an unnest function by using the column analyzer to check ACL on all input expressions for an unnest function.

[rschlussel]

can we do this now by checking analysis.getType(expression)?

[kevintang2022]

Yes I can check for Map type

[rschlussel]

thanks for this fix!

[rschlussel]

technically it might be possible to do even better by adding keys and values expressions separately in case only one is used, but that seems complicated to do, so probably not worth it.

[kevintang2022]

Yes I think the case you're referring to is
Unnest(MAP(array_agg(a), array_agg(b))) t (c1, c2)

If only c2 is used, then we should only count b as utilized, but this change would count both a and b as utilized

[rschlussel]

there's a merge conflict now. you'll need to rebase.