Add a statistic around the number of files being checkpointed. The statistic should reset to 0 at the checkpoint start, then be incremented as a new file gets picked up for checkpointing. It will help us obtain the following information:

• How many files are a part of a checkpoint
• Exactly when a file gets picked up for the checkpoint, and hence the duration a single file is being checkpointed. This would be helpful in mapping the eviction statistics to the individual file checkpoint and answer questions like are there many files to checkpoint, are there a smaller number of files that are taking a long time, is there a single file that takes a long time etc. - Since eviction can not run in parallel to checkpoint on a file - it could have come handy in debugging a few customer cases of late where I suspected checkpoint is running on large collections.