Better testing for Interactive mode

---
Seems like things break a lot in interactive and we don't realize it until people try it. Interactive needs unique, thorough test, that test the flow of information and function calls, not just tests for specific function calls.

Things to test more thoroughly:
- [x] gathering commands from help dump
- [ ] writing command table information into help dump
- [x] success in instantiating a completer, and its validity in processing command
- [ ] use of --progress
- [ ] scoping and ability to run correct commands in different scopes
- [ ] success in storing/reading from history

This is not an exhaustive list and will be ongoing work
### Environment summary

Install Method (e.g. pip, interactive script, apt-get, Docker, MSI, edge build) / CLI version (`az --version`) / OS version / Shell Type (e.g. bash, cmd.exe, Bash on Windows)

```
```