Throw error on missing environment

[sgomes]

This PR introduces a check for missing environment configuration when explicitly providing an environment name, either by including it in opts or by providing it via the BROWSERSLIST_ENV environment variable.

It now throws an explicit error, whereas before it silently used the config defaults.

As discussed in #559, this is a breaking change.

Fixes #559.

[Semigradsky]

What if add an option about generate or not error in this case?
It can be not breaking change if this option will be disabled by default.
Tools which used browserslist can enable option in theirs major releases but browserslist will remain in 4.x version.

[sgomes]

Thanks for taking a look at this PR, @Semigradsky! I'm not sure I follow your suggestion, however. The idea is to only enable this in browserslist 5.x, since it's a breaking change. As such, any tools that use browserslist 4.x will keep the old behaviour, and only tools that get updated to browserslist 5.x will get the new behaviour. This is the correct use of a major version number, in my opinion.

Adding an option seems to me like a bad idea, because it could lead to different browserslist 5.x-enabled tools behaving differently for the same browserslist config and version.

Perhaps I'm misunderstanding your suggestion?

[Semigradsky]

Don't worry, this is just my suggestion on how to release it in the near future)
Releasing a major version of browserslist can be painful.

[ai]

@Semigradsky anyway we need major release for other features

[Muradalsaaideh]

This PR introduces a check for missing environment configuration when explicitly providing an environment name, either by including it in opts or by providing it via the BROWSERSLIST_ENV environment variable.

It now throws an explicit error, whereas before it silently used the config defaults.

As discussed in #559, this is a breaking change.

Fixes #559.

[Muradalsaaideh]

This PR introduces a check for missing environment configuration when explicitly providing an environment name, either by including it in opts or by providing it via the BROWSERSLIST_ENV environment variable.

It now throws an explicit error, whereas before it silently used the config defaults.

As discussed in #559, this is a breaking change.

Fixes #559.

111e1f6