feat(ha): enhance MySQL InnoDB Cluster & Group Replication diagnostics

Summary & Motivation

Enhance MySQLTuner diagnostics and analytics for MySQL InnoDB Cluster and Group Replication topologies:

1. Advanced MySQL Router Metrics & Query Categorization: Parse processlist for MySQL Router connections, monitor active vs sleeping router sessions, and detect/warn when write statements (INSERT, UPDATE, DELETE, DDL) are routed to SECONDARY cluster nodes.
2. Dynamic Single-Primary vs Multi-Primary Mode Adaptation: Adapt flow control recommendations and lower certification conflict rollback warning threshold from 5.0% to 2.0% when running in Multi-Primary mode (group_replication_single_primary_mode = OFF).
3. Message Cache Retention Capacity Audit: Cross-reference group_replication_message_cache_size against active network traffic throughput (Bytes_sent + Bytes_received / Uptime) to calculate estimated network isolation retention window and warn if retention < 60 seconds to prevent state transfer (SST) recovery spikes.
4. Topology Status Output: Standardize output for MEMBER_STATE and MEMBER_ROLE across all cluster nodes.

Changes Included

• Updated mysqltuner.pl with Group Replication and Router diagnostic enhancements.
• Added comprehensive unit subtests in tests/unit_ha_cluster.t.
• Validated via E2E lab tests (build/test_ha.sh innodb_cluster).