Select larger components for Q20a

[szarnyasg]

Q20a would be more challenging if it would be selected from large components. However, it seems that our opportunities are limited: even the largest components consist of only ~277k persons.

D create or replace table ksu as select * from read_parquet('factors-sf10000/parquet/raw/composite-merged-fk/sameUniversityConnected/*.parquet');
D select distinct component, count from ksu order by count desc limit 5;
┌───────────┬────────┐
│ Component │ count  │
├───────────┼────────┤
│ 306       │ 277888 │
│ 3206      │ 277610 │
│ 6598      │ 277330 │
│ 4784      │ 277029 │
│ 679       │ 276735 │
└───────────┴────────┘