[ML] Update model_size_stats as soon as an anomaly detection job is opened #121168
Description
When we open a job, the memory tracker will reserve model_memory_limit bytes for this model on an ML node. However, this usage is not reflected in model_size_stats from GET _ml/anomaly_detectors/_stats?filter_path=jobs.model_size_stats until we start the datafeed and the job starts to produce the first results and report model_memory_stats.
This means that between opening the job and starting the datafeed, the reported memory usage in model_size_stats is 0:
{
"model_size_stats": {
"job_id": "my-job-id",
"result_type": "model_size_stats",
"model_bytes": 0,
"total_by_field_count": 0,
"total_over_field_count": 0,
"total_partition_field_count": 0,
"bucket_allocation_failures_count": 0,
"memory_status": "ok",
"categorized_doc_count": 0,
"total_category_count": 0,
"frequent_category_count": 0,
"rare_category_count": 0,
"dead_category_count": 0,
"failed_category_count": 0,
"categorization_status": "ok",
"log_time": 1736525034132
}
}The actually should be something like
{
"model_size_stats": {
"job_id": "my-job-id",
"result_type": "model_size_stats",
"model_bytes": 0,
"model_bytes_memory_limit": 11534336,
"assignment_memory_basis": "model_memory_limit",
"total_by_field_count": 0,
"total_over_field_count": 0,
"total_partition_field_count": 0,
"bucket_allocation_failures_count": 0,
"memory_status": "ok",
"categorized_doc_count": 0,
"total_category_count": 0,
"frequent_category_count": 0,
"rare_category_count": 0,
"dead_category_count": 0,
"failed_category_count": 0,
"categorization_status": "ok",
"log_time": 1736525034132
}
}