Context

Before entering the critical section, checkpoint optionally waits in checkpoint wait reduce dirty cache (src/checkpoint/checkpoint_txn.c) while eviction reduces cache dirty toward eviction checkpoint target. That loop currently has no WT VERB CHECKPOINT PROGRESS output.

Prepare and write-phase progress messages exist elsewhere, but scrub wait is silent. Reusing the write-phase helper would be misleading because write/walk/file counters are still near zero:

Checkpoint has been running for N seconds, wrote X pages (Y MB), walked Z pages and checkpointed F files

Motivation

A long scrub wait looks like a hang or a quiet write phase. Engineers diagnosing slow checkpoints need a heartbeat that identifies scrub state and whether dirty is moving toward the target.

Suggestions

1. From the scrub wait loop (after each sleep), emit a scrub-specific progress message gated on WT PROGRESS MSG PERIOD.
2. Add a dedicated helper (e.g. checkpoint scrub progress); do not reuse the write-phase checkpoint progress text.
3. Include elapsed time, current dirty %, scrub target %, and bytes written while waiting. Suggested form:

   Checkpoint scrub has been running for N seconds, cache dirty X% (target Y%), wrote Z MB while waiting
   

1. Keep this separate from prepare and write-phase messages so the phase is obvious in the log.

Examples

Desired:

[WT_VERB_CHECKPOINT_PROGRESS]: Checkpoint scrub has been running for 40 seconds, cache dirty 28% (target 5%), wrote 120 MB while waiting