command.py: use argparse subparsers

[himdel]

This is trying to do a similar thing to #30 , but reducing the scope to just moving all custom param parsing to argparse subparsers with no syntax changes and no DRYing up the actual commands yet.

The effect should be better help, more reliable failures with unexpected param counts, and ability to add named arguments easily.

(help description can be added to each kind, op, subop and arg, but it can be done gradually, mostly leaving out of this PR; It looks #30 has a lot more help text ready.)

(args.function is also out of scope, for now just having the subparser dest= to the original arg name (kind, operation, subop))
(args.rest is gone now)

Renaming args.username & args.password to args.auth_username and args.auth_password, to prevent clashes with the username and password positional arguments for user create. (version would also conflict but main uses action="version", not action="store")

Also adding --debug and --version params, and admin default values for username and password.

[himdel]

Seems to work without breaking stuff for CI .. ansible/ansible-hub-ui#2096 (https://github.com/ansible/ansible-hub-ui/runs/6553850868?check_suite_focus=true)

[himdel]

Cc @hendersonreed this is my attempt to revive the command.py refactor by doing the minimum needed to get argparse parsing all the options instead of us splitting args.rest manually. WDYT? :)

(If this makes sense, we can follow up with taking the command cleanup parts from #30, adding more help strings, etc.)

[hendersonreed]

I think this is necessary, and all current QE environments using galaxykit are using an older version. If this does break them, we should update them when it comes time (but I don't think it does.)

IMO this is safe to merge to master for sure - all QE use is with versions that are released in PyPI.