Master should not fragment replication events based on slave_max_allowed_packet

MDEV-32570: Fragment ROW replication events larger than max_packet_size

Currently,When the master encounters a Rows_log_event larger than the slave’s configured slave_max_allowed_packet, it fragments that row event into multiple partial events—each fitting within the allowed size.

However, slave_max_allowed_packet is strictly a slave-side setting and is generally not known to the master. In a typical topology, the master has no reliable way to determine the value configured on one or more replicas.

Fragmentation of replication events should be determined solely by master-side configuration, not by variables defined on the replica.

To address this, the fragmentation threshold should be based on a variable available on the master, for example:

1. Reuse the existing max_allowed_packet setting on the master, or
2. Introduce a new dedicated master-side variable specifically for controlling replication event fragmentation.

Additionally, a requirement should be documented and enforced that:

The value of slave_max_allowed_packet on the slave must be greater than or equal to the master’s fragmentation threshold (max_allowed_packet or the new variable), so that large updates using row-based replication do not cause replication failures.