Support native bit_and / bit_or / bit_xor aggregate

[zhuxiangyi]

Background

Auron does not implement the bitwise aggregates bit_and / bit_or / bit_xor natively, so they fall back to the generic UDAF path (a JNI call back into the JVM) and lose vectorized acceleration. These aggregates are commonly used for flag/bitmask rollups (e.g. bit_or to union feature flags across rows, bit_and for common flags) and parity/checksum use cases.

Proposal

Add native bit_and / bit_or / bit_xor aggregates:

1. Implement a generic AggBitwise<P> in datafusion-ext-plans (agg/bitwise.rs), parameterized by the bitwise operator, with type aliases AggBitAnd / AggBitOr / AggBitXor. The accumulator is a single column of the input type: the first non-null value initializes the slot and every subsequent value is folded in with the operator. The operators are associative and commutative, so the result is independent of the visiting/merge order, and null inputs are skipped (an all-null group yields null). Only integral inputs (Int8/Int16/Int32/Int64) are accepted.
2. Wire the three functions through the AggFunction enum, create_agg, the protobuf contract (BIT_AND / BIT_OR / BIT_XOR), the protobuf::AggFunction -> AggFunction conversion, and the window-aggregate mapping.
3. Add the BitAndAgg / BitOrAgg / BitXorAgg expression conversions in NativeConverters.convertAggregateExpr.
4. Declare the native aggregate buffer schema in NativeAggBase.computeNativeAggBufferDataTypes (Seq(dataType), a single column) so the partial -> shuffle -> final buffer schema matches the native side.

Scope / Non-goals

• Integral inputs only (byte / short / int / long), matching Spark's BitAndAgg / BitOrAgg / BitXorAgg.
• Window aggregate (bit_* over a window) reuses the same AggFunction.

Tests

• Rust unit test agg_exec::test::test_agg_bitwise: partial -> final two-phase aggregation over a nullable integer column, verifying bit_and / bit_or / bit_xor (including null skipping).
• Scala end-to-end test in AuronDataFrameAggregateSuite ("native bit_and / bit_or / bit_xor aggregate", spark34 + spark35): a grouped aggregate exercising the full partial -> shuffle -> final native path (including an all-null group), asserting correct values and that the plan offloads to NativeAggBase.

[zhuxiangyi]

PR: #2363