feat: implementation of unified ORM schema synchronization system via Schema AST

[rogerpadilla]

Summary

This PR implements a comprehensive, graph-based schema synchronization system for UQL. By introducing Schema AST as the core foundation, we now have a unified, dialect-agnostic representation of database schemas where relationships and indexes are first-class citizens. This architecture enables reliable schema diffing, smart relation detection, and powerful synchronization workflows.

Key Features

• Schema AST Foundation: A unified graph representation of tables, columns, relationships, and indexes. Supports topological sorting for deterministic CREATE/DROP ordering and built-in circular dependency detection.
• Unified SQL Generation: Refactored the multi-class generator hierarchy into a single, high-performance SqlSchemaGenerator parameterized by dialect-specific configurations.
• Type-Safe Migration Builder: A new fluent API for defining migrations with complete type safety and automatic foreign key handling.
• Database-First Codegen: Ability to generate clean, decorated TypeScript entities directly from an existing database schema, including intelligent relation inference (e.g., detecting junction tables for Many-to-Many).
• Schema Sync & Drift Detection:
  • autoSync: Safely push entity changes to the database.
  • DriftDetector: Identify and report discrepancies between entities and production storage.
  • EntityMerger: Safely update existing entity files with new database changes using the TypeScript compiler API.
• Standardized Introspection: All SQL database introspection now produces a common SchemaAST, significantly reducing implementation logic in specific drivers.

Documentation & CLI

Updated CLI with new powerful commands:

• generate:from-db: Create entities from an existing database.
• sync:entities: Keep entities in sync with database changes.
• drift:check: Verify if the database schema matches your entity definitions.

Quality & Refactoring

• Removed dozens of as any type assertions across migration and schema modules.
• Refactored BaseSqlIntrospector to reduce cognitive complexity.
• 100% clean biome check and tsc --noEmit validation.
• 602 tests passing with high coverage across all 8 supported database engines.

Closes #80