feat: add ReplicationPlugin for external database data replication

[tarai-dl]

Description

Implements pull-based data replication from external databases (Postgres, MySQL, etc.) into StarbaseDB internal SQLite Durable Object.

This addresses the feature request in #72 by providing a plugin that can periodically pull data from an external source (e.g., Supabase Postgres) into the internal DO SQLite, allowing the instance to serve as a close-to-edge replica.

Features

• Pull-based replication: Periodically fetch data from external sources using Durable Object Alarms
• Incremental sync: Track progress using a configurable tracking column (e.g., id, created_at, updated_at)
• Table filtering: Choose which tables and columns to replicate
• Configurable intervals: Set custom sync intervals (default: 1 minute)
• Conflict strategies: Handle duplicate rows with replace, ignore, or update
• State tracking: Monitor sync progress per table with tmp_replication_state table
• Sync logging: Track sync history with tmp_replication_log table
• REST API: Manage and monitor replication via HTTP endpoints

API Endpoints

Method	Path	Description
GET	/replication/status	Get sync status for all tables
GET	/replication/logs?limit=50	Get recent sync logs
GET	/replication/tables	List configured tables
POST	/replication/sync	Trigger manual sync

Usage Example

import { ReplicationPlugin } from '../plugins/replication'

const replicationPlugin = new ReplicationPlugin({
    intervalMs: 30000, // Every 30 seconds
    batchSize: 1000,
    conflictStrategy: 'replace',
    tables: [
        {
            sourceTable: 'external_users',
            targetTable: 'users',
            trackingColumn: 'updated_at',
            columns: ['id', 'name', 'email', 'updated_at'],
        },
        {
            sourceTable: 'orders',
            trackingColumn: 'created_at',
            filter: "status = 'completed'",
        },
    ],
})

// Add to plugins array
const plugins = [
    // ... other plugins
    replicationPlugin,
]

Implementation Details

• Uses Durable Object Alarms for scheduling (no external cron service needed)
• Creates tmp_replication_state and tmp_replication_log tables for tracking
• Supports incremental sync via tracking column comparison
• Auto-creates target tables from external schema on first sync
• Comprehensive error handling with state persistence

Closes #72