[Iceberg] Enable affinity scheduling on file sections

[ZacBlanco]

Description

This change moves the affinity scheduling file section size
configuration from HiveClientConfig and HiveSessionProperties
to HiveCommonClientConfig and HiveCommonSessionProperties so
that the iceberg connector can benefit from this scheduling
strategy when tables have a small number of files but a large
number of splits.

Motivation and Context

On tables with a small number of large files, queries may perform poorly due to the distribution in split scheduling being skewed. This is more likely to occur when there is a limited number of values being hashed to determine the preferred nodes to schedule to. By changing the identifier used for selecting the preferred nodes we increase the probability that the splits are scheduled more evenly across the cluster.

Impact

• Hive-specific configuration moved to common configuration.

Test Plan

Added a unit test to verify that the number of unique identifiers changes as we scale up the file section size

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

== RELEASE NOTES ==

Iceberg Connector Changes
* Add support for the ``hive.affinity-scheduling-file-section-size`` configuration property and ``affinity_scheduling_file_section_size`` session property.

[steveburnett]

LGTM! (docs)

Pull branch, local doc build, looks good. Thanks!

[yingsu00]

Mostly looks good. One minor correction: "splits not being scheduled to enough nodes" : It's not necessarily they were not scheduled to enough nodes, but in general it had more skew than Hive, even when the splits were scheduled to the same number of nodes. Scheduling to less nodes happened non-determistically when I ran the queries multiple times. More than half times they did were scheduled to all nodes, but even in such cases the load was not as balanced as Hive.

[ZacBlanco]

Thanks for the feedback @yingsu00 - I updated the PR description to be a bit more accurate

[aaneja]

Should we make TARGET_SPLIT_SIZE a DataSize instead of refering to bytes ? We do this for MAX_SPLIT_SIZE. Can punt this to a new PR

[yingsu00]

@aaneja Good catch!
@ZacBlanco Could you please address Anant's comment in this PR? I'll merge once it's done.

[ZacBlanco]

This was done intentionally because the Iceberg library expects this configuration in bytes directly on the "read.target.split-size" table property. Iceberg has no notion of "DataSize" or unit-based shorthand for byte-typed values. In order to make setting the values for the table and session property consistent for the target split size I opted to not make this a DataSize-typed property.