Profile behavior behaves badly when unsuccessful

[waynew]

See #351 for more info, but generally speaking there are two behaviors when using profiles that are just terrible.

<snip>
  File "/home/wayne/mine/review-saltext/src/saltext/vmware/utils/connect.py", line 45, in get_config
    credentials = conf[profile]
KeyError: 'fnord'

A huge stack trace appears when doing something like salt-call vmware_vm.list profile=fnord and the profile doesn't exist.

We could pretty easily catch a KeyError here and then look for profiles that might match, and provide a useful error/exit like "Profile 'fnord' does not exist. Perhaps you meant X?"

Alternatively, if you don't have any config data in the base and only have profiles, and do not provide a profile, you get this:

<snip>
  File "/home/wayne/mine/review-saltext/src/saltext/vmware/utils/connect.py", line 134, in get_service_instance
    config = get_config(config=config, profile=profile, esxi_host=esxi_host)
  File "/home/wayne/mine/review-saltext/src/saltext/vmware/utils/connect.py", line 60, in get_config
    raise ValueError("Cannot create service instance, VMware credentials incomplete.")
ValueError: Cannot create service instance, VMware credentials incomplete.

Also super useless.

Basically what we should do in this case is check if there's a host/user/pass, and if any of those things are missing then we should provide a useful message such as:

No credentials found in the base profile, and no profiles found.

Or

No credentials found in the base profile, but profiles <a, b, c> exist. Are you missing profile=a?

Something to that effect.

[jgangel]

This is the behavior when using connect.get_service_instance but a different handling/error is done when using connect.request which is used by other modules. Not sure how profiles are supposed to work, is there any guide for using them?