fix: data export - replace O(n²) org unit descendants cross-join with path index scan [DHIS2-21490]

[jbee]

Summary

While testing I noticed that data export was unreasonably slow. I found that the SQL to expand direct OU matches to include all children was the cause. I ended up discussing this with Jason who came up with the SQL in this PR. The same query that took 2 minutes or so before now was instant.

Query plan

before:

Nested Loop Left Join ... (actual time=2.156..681.690 rows=1333 loops=1)
  Join Filter: ((ou_1.path)::text ~~ ((parent_ou.path)::text || '%'::text))
  Rows Removed by Join Filter: 1,771,738

Query plan here: https://explain.depesz.com/s/eK52 (638 ms)

after:

Nested Loop (actual time=0.040..1.049 rows=1333 loops=1)
  Join Filter: ou_1.path ~~ root.path || '%'
  ->  Seq Scan on organisationunit ou_1  (1333 rows)
  ->  Materialize (rows=1 loops=1333)

Query plan here: https://explain.depesz.com/s/s5Voe (8 ms)

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud