Optimize KMS key reconciliation process

[gc-tejas]

This pull request enhances the KMS key reconciliation process by optimizing how missing data keys are detected and handled across multiple scopes and purposes. The main improvement is the introduction of a new repository method that efficiently identifies which scopes lack specific data keys, enabling batch processing and reducing unnecessary lookups. Comprehensive tests have also been added to ensure correctness, including edge cases and large batches. This addresses ICU-17643 and ICU-17641

Key improvements to key reconciliation logic:

• Added a new repository method, ListScopesMissingDataKey, which efficiently returns a map of missing data keys for each purpose across a list of scopes, supporting batching for large requests.
• Refactored the Kms.ReconcileKeys method to use the new repository method, allowing it to only process scopes and purposes that are actually missing data keys, improving performance and clarity.

Testing improvements:

• Added comprehensive tests for ListScopesMissingDataKey, covering cases such as missing keys, missing root keys, all keys present, mixed scenarios, and large batch operations.
• Extended key reconciliation tests to cover scenarios with multiple scopes and mixed missing keys, as well as large batch processing.

Supporting changes:

• Added query, postgresScopesMissingDataKeyQuery sqliteScopesMissingDataKeyQuery, to efficiently find missing data keys for given scopes and purposes.
• Added necessary imports and minor code cleanups to support new functionality and tests.

[hashicorp-cla-app]

All committers have signed the CLA.