feat: Added an ability to pass a replication count via the CLI

[Oloremo]

Just adding an ability to pass a replication count for upload, default is preserved.

[Oloremo]

Hm, the CI errors seems unrelated to the PR:

 pip install .[avro] coverage mock
before_script.1
0.35s$ export HADOOP_HOME="$(./scripts/hadoop.sh download)"
before_script.2
0.02s$ export HADOOP_CONF_DIR="$(./scripts/hadoop.sh config)"
0.01s$ ./scripts/hadoop.sh start
./scripts/hadoop.sh: line 67: /bin/hdfs: No such file or directory
./scripts/hadoop.sh: line 68: /sbin/hadoop-daemon.sh: No such file or directory
The command "./scripts/hadoop.sh start" failed and exited with 1 during .
Your build has been stopped.

[mtth]

Apologies for the delayed review @Oloremo. Two minor comments and this should be good to go.

[mtth]

Let's shorten the name to fit within the left hand side. For example:

-r COUNT --replication=COUNT

Can you also move it before -s to keep the options sorted?

[mtth]

The current code doesn't set the default. Can you keep this behavior when the option isn't set? The cluster might override the default site-wide.

[Oloremo]

I missed your reply and switched to other stuff. :-(

The current code doesn't set the default.

I'm super confused with that docs-as-a-config thing, but as I understand, [default: 3] sets the defaults?

Can you keep this behavior when the option isn't set? The cluster might override the default site-wide.

Not sure I follow. The cluster does have site-wide settings but the client can override them. So if you don't pass that fro the client perspective - cluster defaults will be used.

[mtth]

The current code (before this PR) doesn't set the default on the client, so the cluster's default will be used. This is a nice property which would be lost after this change since the client now sets a default of 3. Instead the client should only set replication when explicitly specified by the user.