ember try:ember seems to be doing a bit too much

I [opened a PR](https://github.com/ember-cli/ember-try/pull/850) a little while ago and I noticed that some of the tests were failing. Digging into it a little bit, it seems like the failure is to do with the `ember try:ember` command.

Looking at [the results printout](https://github.com/ember-cli/ember-try/runs/6424723584?check_suite_focus=true#step:6:877) for the scenario `./node_modules/.bin/ember try:ember '3.2.0' --skip-cleanup=true` it looks like it's running way more tests than just `ember-source@3.2.0` 🤔

Looking at the output it is running tests for

- 4.5.0-beta.1.beta
- 4.6.0-alpha.1.canary
- 4.4.0-release
- 3.2.0
- 2.12.2
- 2.16.4

The first surprising thing to me is that it runs **anything** other than `3.2.0` 🤔 I guess I could come to peace with the fact that it ran `3.2.0` as well as the scenarios that were defined in the ember-try config. But I can't see why this command would result in the beta, alpha, and release versions getting run 🤔

I suspect this is just a bug in the implementation. If you can confirm what it's intended to test when you run `ember try:ember '3.2.0'` I will see if I can update the tests and implementation to correspond with this.

Alternatively since [I think very few people are actually](https://emberobserver.com/code-search?codeQuery=try%3Aember) using the `try:ember` command, maybe we just deprecate it and remove the test?

What do you think?