fix: task log pagination

[jason-lynch]

Summary

Fixes a bug where the last_entry_id is empty when after_entry_id is the last entry in the log. Instead of performing an exclusive get in Etcd, we perform an inclusive get and then filter out the entry with entry_id == after_entry_id in the service layer.

Testing

To test this change:

# perform an async operation, such as creating a db
# note that the follow-task helper should continue to work as expected
cp1-req create-database < ./.scratch/requests/db.json | cp-follow-task

# once the task is complete, list tasks
cp1-req list-database-tasks <db id>

# get the task logs
cp1-req get-database-task-log <db id> <task id>

# take the last entry id from the previous response and use it as the
# after-entry-id argument
cp1-req get-database-task-log <db id> <task id> --after-entry-id <entry id>

# note that the entries list is empty, but the last_entry_id is still
# populated.

# if needed, you can further validate that the correct entries and number of
# entries is returned by comparing with the contents in etcd:
cp-etcdctl get --sort-by key --prefix /task_log_messages/<db id>/<task id> --keys-only | sort -u | tail -n +2

PLAT-367