Reenable float aggregation pushdown for duckdb except for min()

[myrrc]

Float aggregate pushdown was disabled in #9069 because of nan handling. For max,avg,sum,count, and first Vortex's behaviour
with include_nans() matches duckdb's so we can enable it.
Duckdb's min() NaN ordering, however, is incompatible so we still forbid
pushdown.

Forbid i64/u64 sum pushdown since an overflow/underflow in Vortex returns NULLs but in Duckdb it returns i128/u128.

Change vortex benchmark to use i32 on columns to measure sum pushdown.

Resolves: #9085

[codspeed]

Merging this PR will not alter performance

✅ 1842 untouched benchmarks
⏩ 55 skipped benchmarks¹

---

_{Comparing myrrc/duckdb-reenable-float-aggregations (4aba56f) with develop (e4898d6)}

Footnotes

1. 55 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩

[robert3005]

fwiw the default is to skip nans so I think we can push it down?

[myrrc]

If we include nans, min([nan,1,2]) = NaN, and duckdb says 1, because -NaN is treated same as NaN.
If we skip nan, min([nan,nan]) = NULL, and we need NaN.

[robert3005]

thanks for spelling this out. I missed some of the detail. DuckDB semantics are truly... special.