Add composite bucket aggregation support

[varun-st]

Description

Adds support for composite bucket aggregations to the DSL query executor plugin, enabling multi-dimensional grouping with efficient cursor-based pagination.

Changes

Core Implementation

CompositeGrouping
• Container for multiple source groupings (terms, histogram, date_histogram)
• Implements cursor-based pagination via buildAfterKeyFilter()
• Generates lexicographic comparison filters: (col1 > v1) OR (col1 = v1 AND col2 > v2) OR ...

CompositeBucketTranslator
• Converts composite aggregations to internal representation
• Extracts per-source sort direction and missing value ordering
• Builds InternalComposite response with proper bucket keys

GroupingUtils
• flatten(): Flattens composite groupings into individual sources
• hasExpressionGrouping(): Detects expression-based groupings (histogram, date_histogram)

AggregateConverter & AggregationMetadataBuilder
• Extended to handle composite groupings
• Injects after-key filter for pagination before GROUP BY
• Processes each source grouping for metadata generation

Key Features

✅ Multi-dimensional grouping (terms + histogram + date_histogram)
✅ Cursor-based pagination with after parameter
✅ Per-source sort direction (ASC/DESC)
✅ Per-source missing value ordering (FIRST/LAST)
✅ Lexicographic ordering for consistent pagination
✅ Nested sub-aggregations support

SQL Translation

Composite aggregations translate to a single SQL query with:
• Multi-column GROUP BY for all sources
• WHERE clause for cursor-based filtering (when after is provided)
• Computed expressions for histogram/date_histogram sources

Example:
json
{
"composite": {
"sources": [
{ "category": { "terms": { "field": "category" } } },
{ "price_bucket": { "histogram": { "field": "price", "interval": 100 } } }
],
"after": { "category": "Electronics", "price_bucket": 200 }
}
}

Generated SQL:
sql
SELECT
category,
FLOOR((price - 0) / 100) * 100 AS price_bucket,
COUNT(*) AS _count
FROM table
WHERE (category > 'Electronics')
OR (category = 'Electronics' AND FLOOR((price - 0) / 100) * 100 > 200)
GROUP BY category, FLOOR((price - 0) / 100) * 100
ORDER BY category ASC, FLOOR((price - 0) / 100) * 100 ASC

Testing

Unit Tests:
• CompositeGroupingTests: Field name extraction, after-key filter generation
• CompositeBucketTranslatorTests: Source conversion, ordering configuration
• GroupingUtilsTests: Flattening and expression detection

Integration Tests:
• Basic composite with single/multiple sources
• Pagination with no duplicate results across pages
• Multi-dimensional pagination with lexicographic ordering
• Custom sort order and missing value handling