Key Changes:

RLookup Synchronization:

• Use new RLookup_CloneInto() to copy upstream lookup into tail pipeline
• Aligns field indices (dstidx, svidx) between search subquery and tail pipeline

Pipeline Refactoring:

• Moved Pipeline_BuildAggregationPart case PLN_T_LOAD logic into a separate function
• Reused this new function in both Pipeline_BuildAggregationPart and HybridRequest_BuildPipeline
• In HybridRequest_BuildPipeline, used a cloned load step for this logic

Test Changes:

• Refactor AddLoadStepToPlan to create unprocessed LOAD steps (with ArgsCursor instead of pre-populated keys)
• Tests now verify that BuildPipeline correctly processes LOAD steps from unprocessed to processed state