The load_and_stream nemesis must use unique and temporary table that gets deleted afterwards

[vponomaryov]

Scenario: 5-dc/60 nodes DB cluster setup affected by the https://github.com/scylladb/scylla-enterprise/issues/4784
Argus: scylla-staging/valerii/vp-longevity-aws-custom-d2-workload1-multidc-big#19

The load_and_stream nemesis has following problems:

It doesn't delete the data and table it created in the end of it's execution.

It affects DB cluster in further nemesis scenarios like the destroy_data_then_repair.
5-dc big DB cluster performs slow repairs (https://github.com/scylladb/scylla-enterprise/issues/4784) even without additional tables.
But running after the load_and_stream nemesis it becomes even more slower just because there are more data than expected.
Significantly more.

So, need to update the load_and_stream in the following way:

• Make it use unique and temporary table for populating data, not standard C-S table - standard1.
• Make it delete data and table in the end of it's execution.