fixed broken behavior of buffalo test -m

[sio4]

Not sure when the issue started, but buffalo test -m TEST was not working as expected. The bug was reported as #180 recently (transferred from the buffalo repository).

This fix fixes the following things:

• removed unnecessary .buffalo.dev.yml existence checking on buffalo dev command. The logic should not reach there since the rootCmd checks that before executing subcommands.
• fixed wrong anywhereCommands checking. anywhereCommands means they can be executed outside app root.
• fixed not to register plugins as anywhereCommands. (maybe related to the above miss-checking?)
• fixed buffalo test -m handles argument properly especially with testify.

fixes #180

[paganotoni]

This -m flag is something I've been thinking about while working on the new CLI architecture. I've been thinking if this is something standard in Go or something that Testify requires and how do we write our CLI so we are not dependent on Testify.

[sio4]

This -m flag is something I've been thinking about while working on the new CLI architecture. I've been thinking if this is something standard in Go or something that Testify requires and how do we write our CLI so we are not dependent on Testify.

In our code, the -m works as two ways.

• if the code can detect testify-enabled on a module, the -testify.m TestCase flag is used instead -m itself.
• if the module to be tested is not testify-enabled, the -run TestCase flag is used instead of -m itself.

So what I found for -m is a kind of meta flag to be transformed into the standard -run or testify style and I think this method is useful. (Actually, the scaffolded app supports testify)

[paganotoni]

Was taking the time to review how this impacted the v1 version of all this. Thanks for adding the comments @sio4.