Fix signature for Kernel.system.

[reitermarkus]

Motivation

Kernel.system has some functionality which is not covered by the current signature at all.

Currently, out of these

system('echo')
system('echo', err: File::NULL)
system('echo', out: :err)
system('echo', 'hello')
system('echo', 'hello', err: File::NULL)
system('echo', 'hello', out: :err)
system(['echo', 'echo'])
system(['echo', 'echo'], 'hello')
system(['echo', 'echo'], out: :err)
system(['echo', 'echo'], 'hello', out: :err)
system(({'VAR' => 'VAL'}), 'echo')
system(({'VAR' => 'VAL'}), 'echo', out: :err)
system(({'VAR' => 'VAL'}), ['echo', 'echo'])
system(({'VAR' => 'VAL'}), ['echo', 'echo'], out: :err)
system(({'VAR' => 'VAL'}), ['echo', 'echo'], 'hello')
system(({'VAR' => 'VAL'}), ['echo', 'echo'], 'hello', out: :err)

only

system('echo')
system('echo', 'hello')

type-check. The fixed signature should cover all of these, however it also accepts some configurations which are wrong, i.e.

system('echo', ['echo', 'echo'])
system(['echo', 'echo'], ['hello', 'hello'])

I'm not quite sure if there is a way to make the second argument depend on the first argument, i.e. only allow [String, String] as the second if the first is T::Hash[String, T.nilable(String)].

Test plan

n/a

[elliottt]

Could you add some tests for calling Kernel.system with this new signature in test/testdata/rbi/kernel.rb?

[elliottt]

The signature change makes sense, but adding some tests to make sure that expected uses are possible would be great.

[elliottt]

We have a policy of testing changes to Sorbet against Stripe's codebase before
merging them. I've kicked off a test run for the current PR. When the build
finishes, I'll share with you whether or how it failed. Thanks!

Stripe employees can see the build result here:

→ https://go/builds/bui_IS5iftBlSiD2Rm

[elliottt]

Thanks!

We should hold off on merging this until the stripe migration PR has also been merged.