[Feature][Metric] Custom SQL support export error data

[xxzuo]

Purpose of this pull request

Summary

This PR adds a new custom_count_sql metric that allows users to define custom SQL for invalid/error data extraction and automatically calculates the invalid row count from that SQL.

It also refactors custom SQL table discovery so metric plugins can define which SQL field should be used for metadata/table extraction, instead of hard-coding actual_aggregate_sql.

Changes

• Add the custom_count_sql metric plugin.
• Register the new metric plugin through META-INF/services.
• Add invalidate_items_sql as a metric configuration key.
• Add UI support for SQL-style metric config fields using TextArea.
• Add localized placeholders for invalidate_items_sql.
• Add SqlMetric#getTableDiscoverySql and SqlMetric#setTableDiscoverySql.
• Update Local, Spark, and Flink engine config builders to use the new table discovery API.
• Update JobServiceImpl to resolve metadata using the metric-provided table discovery SQL.
• Make CustomCountSql:
  • discover tables from invalidate_items_sql
  • generate invalid item output from the custom SQL
  • count invalid rows from ${invalidate_items_table} for normal engine flow
  • count directly from the custom SQL for Local direct execution

close #560

Check list

• Code changed are covered with tests, or it does not need tests for reason:
• Change does not need document change