Feature: Distributed COPY INTO

[yufan022]

Summary

Should we support distributed copy into from s3 distributed=true in the future?

If we use copy into from s3 purge=true to import data simultaneously in many query-node, we must very careful to set s3-prefix or pattern so that different nodes use different directories to avoid repeatedly importing.

like this:

#query1
copy into xx from 's3://query1/' pattern ='.*[.]tsv' purge=true;
#query2
copy into xx from 's3://query2/' pattern ='.*[.]tsv' purge=true;
#query3
copy into xx from 's3://query3/' pattern ='.*[.]tsv' purge=true;

If we support distributed copy into, things will be easier.

we just need to use:

copy into xx from 's3://import/' pattern ='.*[.]tsv' purge=true distributed=true;

[Xuanwo]

#6395 will make COPY run in a cluster so that copy will be distributed automatically.

Also related to #8128

[PsiACE]

#11943