Description: To accurately measure and validate performance improvements to the reconfigure step during checkpoint pickup, we need a reliable, isolated reproducer. This task involves building a custom WT workgen workload to simulate the exact leader/follower disaggregated storage startup performance at high scale.

The script must run in two isolated phases against a single WiredTiger home directory:

• Phase 1 (Leader): Opens as leader, creates the configured number of layered tables, takes a checkpoint, and closes.
• Phase 2 (Follower): Opens as a follower. It must explicitly time wiredtiger_open without picking up the checkpoint first, and then separately time reconfigure(checkpoint_meta=...) to measure the pure checkpoint pickup cost.
• Phase 3: Warm up 20% of the cache and explicitely time it

Definition of Done:

• Workgen script successfully runs both Phase 1 (Leader setup) and Phase 2 (Follower pickup).
• The script accurately times the reconfigure step in isolation.
• Benchmarks can be cleanly executed for 100k, 250k, and 500k table scale points.