Add WarehouseLambda core with S3 WarehouseDumper

[jwils]

This commit implements the core S3 data export functionality for the
elasticgraph-warehouse_lambda gem, including the main WarehouseLambda
class and WarehouseDumper implementation.

Components added:

• WarehouseLambda: Main class that orchestrates indexer, S3 client, and
  warehouse dumper with from_parsed_yaml factory method
• WarehouseDumper: Implements DatastoreIndexingRouter interface to dump
  indexing operations to S3 as gzipped JSONL files
• Configures indexer with WarehouseDumper as the datastore_router to
  redirect all indexing operations to S3 instead of the datastore

Key features:

• Groups operations by GraphQL type and writes separate JSONL files per type
• Generates deterministic S3 keys with versioning, date partitioning, and UUIDs:
  s3://bucket/dumped-data///v//.jsonl.gz
• Structured logging for observability (batch counts, file sizes, S3 keys)
• Extracts data from operation payloads including id, __eg_version, and record fields
• Filters operations to only include primary index updates
• Skips operations with no datastore payload (e.g., delete operations)
• Returns BulkResult with success status for all operations
• Leverages aws-sdk-s3 with SHA256 checksum validation and conditional writes
• S3 client initialization with optional region override from config

Testing:

• 37 comprehensive test examples covering all components
• Tests verify S3 operations, JSONL formatting, compression, and logging
• Uses stubbed AWS S3 client for fast, deterministic tests
• Full RBS type signatures for type safety
• 100% test coverage (from_env marked :nocov: until LambdaFunction added in next commit)

[myronmarston]

Left some more feedback. Most of it is cosmetic...with the exception of the version vs json_schema_version mixup--that one matters!

[myronmarston]

LGTM apart from the build failure. Left a suggestion which should fix it.