[release/6.0] Fix virtualization handling of large dataset length changes. Fixes #37245

[github-actions]

Backport of #37247 to release/6.0

/cc @SteveSandersonMS

Description/Customer Impact

The Virtualize component exhibits pathological behavior in the following case:

• User scrolls to the end of a very long list (e.g., 100000 items)
• The list becomes much shorter (e.g., 100 items)

Instead of jumping back to the new maximum scroll position, it incrementally steps back to it one page at a time. In the above example, if the scroll region is tall enough for 20 items, that would take (100000-100)/20 = 4995 steps. On each step it requests a new data set (possibly causing an HTTP request and DB query) and a UI re-render.

Eventually this does complete, but it's extraordinarily inefficient. It's very trivial to fix this so that we move to the correct final scroll offset in a single step.

Regression?

• Yes
• No

Risk

• High
• Medium
• Low

• Change was verified manually and via a new E2E test
• The code change is trivial: we can detect and correct the invalid scroll offset in a single step. The new logic shouldn't affect any existing case where the scroll offset was valid.

Verification

• Manual (required)
• Automated

Packaging changes reviewed?

• Yes
• No
• N/A

Fixes #37245

[ghost]

Hi @github-actions[bot]. Please make sure you've updated the PR description to use the Shiproom Template. Also, make sure this PR is not marked as a draft and is ready-to-merge.

To learn more about how to prepare a servicing PR click here.