more accurate memory limit calculation for anon-r-seq cases

Signed-off-by: Fengguang Wu <fengguang.wu@intel.com>
3 files changed