RHEL 9 must take action when allocated audit record storage volume reaches 95 percent of the audit record storage capacity.
If action is not taken when storage volume reaches 95 percent utilization, the auditing system may fail when the storage volume reaches capacity.
While "%" is fully compliant for auditd.conf, you must account for how auditd processes it. The percentage-to-megabyte conversion occurs only when the audit daemon starts. If you dynamically resize the underlying logical volume/partition where /var/log/audit resides while the system is running, the daemon will not automatically know. You must reload it to trigger a recalculation.
How this relates to the STIG checklist: STIG vs SCAP benchmark.