1) In the scheduler, a job is an object that is owned by the schema (like any other objects) so it must have a unique name. So you could do something like:

for i in 1 .. 10 loop
  dbms_scheduler.create_job('MYJOB'||i, .... );
end loop;

2) If a scheduler job does not have a repeat interval, it will be dropped just like dbms_job, eg

SQL> begin
  2      dbms_scheduler.create_job (
  3         job_name           =>  'MYJOB',
  4         job_type           =>  'PLSQL_BLOCK',
  5         job_action         =>  'null;',
  6         start_date         =>  sysdate,
  7         enabled            =>  true,
  8         comments           =>  'One off job');
  9  end;
 10  /

PL/SQL procedure successfully completed.

SQL>
SQL> select job_name from user_scheduler_jobs;

JOB_NAME
--------------------------------------------------------------
MYJOB

SQL> -- a short time later
SQL> /

no rows selected

3) There is a lot more flexibility with the scheduler. The concept of a job retrying and ultimately no longer running is still there, but there is a lot of other options as well in terms of monitoring them, or reacting to a failure. For example, you can be emailed on an error, or you can have the scheduler raise an "event" when a job fails, which could trigger other jobs etc.

Check out the details here

https://docs.oracle.com/cd/E18283_01/server.112/e17120/scheduse008.htm