Replace awk -F based csv iteration by IFS + built-in for loop

[kkimurak]

awk -F, splits values by comma but built-in for loop reinterpret each values. This behavior does not allow us to contain internal field separator (such as whitespace) in DB_EXTENSION and DB_NAME.

This PR set environment variable IFS (internal field separator that will affect to shell builtin functions) before / after corresponding for loop so that we can prevent unintended word splitting.

With this change, each element of DB_EXTENSION can include CREATE EXTENSION options.

Example usage (note that the extension vchord will require additional installation process so the container exits after the line):

$ docker run --rm --env 'DB_EXTENSION=btree_gist,pg_trgm,vchord CASCADE' --env 'DB_NAME=test' --env 'DB_USER=test_user' --env 'DB_PASS=password' kkimurak/sameersbn-postgresql:16-test
Initializing datadir...
Initializing certdir...
Initializing logdir...
Initializing rundir...
Setting resolv.conf ACLs...
Initializing database...
Configuring hot standby...
‣ Setting postgresql.conf parameter: wal_level = 'hot_standby'
‣ Setting postgresql.conf parameter: max_wal_senders = '16'
‣ Setting postgresql.conf parameter: checkpoint_segments = '8'
‣ Setting postgresql.conf parameter: wal_keep_segments = '32'
‣ Setting postgresql.conf parameter: hot_standby = 'on'
‣ Setting postgresql.conf parameter: data_directory = '/var/lib/postgresql/16/main'
‣ Setting postgresql.conf parameter: log_directory = '/var/log/postgresql'
‣ Setting postgresql.conf parameter: log_filename = 'postgresql-16-main.log'
‣ Setting postgresql.conf parameter: ssl = 'off'
Creating database user: test_user
Creating database: test...
‣ Loading btree_gist extension...
‣ Loading pg_trgm extension...
‣ Loading vchord CASCADE extension...