1- If you're wanting to do real dication that's accurate (generally that means context-aware), you'll want to steer away from Microsoft's free offerings and go with something like Cortana or Dragon. VoiceAttack's, 'dictation' uses the Windows speech engine and neither one is context-aware.
2- Global profiles are generally used for your common commands between profiles. If you are making profiles for others to use, you'd want to make your profiles self-contained in order to avoid confusion by your users. If you are the only user of your profiles, global profiles are a great way to not have to duplicate commands and pay off big time if you don't have to alter the same command in several profiles.
3- That's something that can be done, however, VoiceAttack provides a lot of different ways to do different things. You'll want to take a look at the VoiceAttack help document, which is located in your installation folder (or can be launched by pressing F1 when VoiceAttack has focus).
4- Not sure what's up there - that's been working for years. Are you using Windows 10?
5- You'll want to add a command that has the action, 'Stop listening' as its first action and another command that has the 'Start Listening' action as its first action.
When VoiceAttack is told to stop listening, it's still, 'listening' for indication to start listening (that is, processing spoken requests) again.
Hope that helps get you started some and welcome to the crew