GeezerVison

Fixes #6226

test_window_agg.py had three negative tests commented out with "TODO: fix this test, it now should pass": RANK, DENSE_RANK and ROW_NUMBER over a constant ORDER BY, which the compiler used to reject outside a TopK pattern. All three are positive tests now.

• RANK() OVER (ORDER BY NULL) and DENSE_RANK() OVER (ORDER BY 1): every row ties, so every row gets rank 1. These only compile since [SQL] Compile RANK when ORDER BY is on a constant #7694, the fix for [SQL] RANK/DENSE_RANK over a constant ORDER BY generates Rust that doesn't compile #7677.
• ROW_NUMBER() OVER (ORDER BY NULL): which row gets which number is unspecified, so the test checks just the numbers 1, 2, 3.

Describe Manual Test Plan

Ran tests/runtime_aggtest/illarg_tests2/main.py against the pipeline-manager:0.372.0 image, which includes #7694, and it passes. To make sure each new view is actually checked, I broke the expected rows of the RANK and DENSE_RANK views one at a time. Each run failed on that view (failed view: illarg_window_rank_order_by_null, ... dense_rank_order_by_const). On the earlier run I did the same for the ROW_NUMBER view. pre-commit run on the file is clean.

Checklist

• Unit tests added/updated
• Integration tests added/updated
• Documentation updated
• Changelog updated

Breaking Changes?

• OpenAPI / REST HTTP API / feldera-types / manager
• Feldera SQL (Syntax, Semantics)
• feldera-sqllib (incl. dependencies fxp, etc.)
• Python SDK
• fda (CLI arguments)
• Adapters (including configuration)
• Storage Format / Checkpoints
• Others (specify)