Add endpoint that fetches completed batches in order

[illusional]

Explain plan: Looks like it's filtering on an index batches_time_completed, so should filter and sort quickly.

-> Limit: 51 row(s)  (cost=308316.58 rows=51) (actual time=0.074..0.520 rows=51 loops=1)
    -> Nested loop inner join  (cost=308316.58 rows=92069) (actual time=0.072..0.515 rows=51 loops=1)
        -> Nested loop left join  (cost=276092.43 rows=92069) (actual time=0.063..0.475 rows=51 loops=1)
            -> Nested loop left join  (cost=180570.84 rows=92069) (actual time=0.049..0.296 rows=51 loops=1)
                -> Nested loop left join  (cost=79778.30 rows=92069) (actual time=0.043..0.191 rows=51 loops=1)
                    -> Filter: (batches.deleted = 0)  (cost=47554.15 rows=92069) (actual time=0.032..0.147 rows=51 loops=1)
                        -> Index range scan on batches using batches_time_completed, with index condition: (batches.time_completed is not null)  (cost=47554.15 rows=184138) (actual time=0.030..0.140 rows=51 loops=1)
                    -> Single-row index lookup on billing_projects using PRIMARY (name=batches.billing_project)  (cost=0.25 rows=1) (actual time=0.000..0.001 rows=1 loops=51)
                -> Single-row index lookup on batches_n_jobs_in_complete_states using PRIMARY (id=batches.id)  (cost=0.99 rows=1) (actual time=0.002..0.002 rows=1 loops=51)
            -> Single-row index lookup on batches_cancelled using PRIMARY (id=batches.id)  (cost=0.94 rows=1) (actual time=0.003..0.003 rows=0 loops=51)
        -> Single-row index lookup on billing_project_users using PRIMARY (billing_project=batches.billing_project, user='$USER')  (cost=0.25 rows=1) (actual time=0.001..0.001 rows=1 loops=51)

[lgruen]

Probably a good idea to move this constant out, as it's repeated in line 467.