Agent2 log monitoring crashes

Initially server process_log_check() server parameter was abused by agent2 to pass internal information through the C code back to the process_value() callback provided by agent2.

With HA address rotation server/port was replaced by address vector, and agent uses the first vector member to get address/port when calling init_persistent_dir_parameter(). On agent2 it will crash.

I don't see how it could be fixed without adding new parameter to process_log_check() and following functions until the callback process_value() is called. This parameter would be used only by agent2.

That would leave init_persistent_dir_parameter() handling, because agent2 does not have the last successful address readily available when processing logs.