Cap the shift amount instead of the result: change WT_MIN(1ULL << retry_count, 1000) to 1ULL << WT_MIN(retry_count, 10) (at wtperf.c:719), so retry_count reaching 64+ can never produce an out-of-range shift.

https://coverity.corp.mongodb.com/#/project-view/24832/10049?selectedIssue=206457