pg_cron logs many errors when default_transaction_read_only on

[serprex]

On managed service default_transaction_read_only is turned on when disk gets full. pg_cron then goes on to generate many logs due to attempting to execute UPDATE. This produces much logging in a scenario where logging would be best kept to a minimum

Ideally pg_cron would either explicitly set the transaction to writable or respect the default & back out