Use multiple flashback query threads for Oracle snapshots [DBZ-4918]

Migrated from [DBZ-4918](https://issues.redhat.com/browse/DBZ-4918)

Hi, I am using debezium to read snapshot from oracle table which has 60 columns and 300000 rows and it's about 220MB. After many times testing I conclude that it should take 14 minutes to finish that snapshot task. I have try to change the option `snapshot.fetch.size` bigger or smaller but got no effection. So it would be greate if you can help to supply some advise to due with this situation.

Here's my configuration of the task.
{code:java}
// code placeholder
connector.class = io.debezium.connector.oracle.OracleConnector
database.tablename.case.insensitive = false
log.mining.strategy = online_catalog
include.schema.changes = false
tombstones.on.delete = false
log.mining.continuous.mine = true
database.history.skip.unparseable.ddl = true
database.history.instance.name = 5fa0ac61-7e7d-4ce8-923c-e704242dc35a
snapshot.fetch.size = 10000
database.user = xxx
database.dbname = xxx
database.url = jdbc:oracle:thin:@//xxx
database.server.name = oracle_logminer
schema.whitelist = xxx
offset.flush.interval.ms = 9223372036854775807
database.password = *****
name = engine
table.include.list = xxx.xxx
snapshot.mode = initial {code}

Start with the OracleConnector snapshot path and the snapshot.fetch.size configuration described in the issue. Determine how Oracle snapshots currently execute and define the changes needed to use multiple flashback query threads; done means the requested snapshot completes with parallel query threads and its behavior is covered by appropriate tests.

Written by the indexing model from the issue text.