We should get rid of Grpc.Core whose maintenance is going to stop

[xlegalles]

See https://grpc.io/blog/grpc-csharp-future/
There are still references to grpc.core in the source code even though most of it uses Grpc.Net.Client. We should get rid of it completely.

[xlegalles]

Hi,
The 1st PR is easy because the ZeebeClient does not use any Grpc.Core code but has only a package reference. But tests require this dependency.
In order to remove this package from tests, we need to refactor unit tests in a second PR. We have 2 options:

• Use a local out-of-process server as it's already done. But to my point of view, this is not UT, this is IT. And we have another set of integration tests using Docker containers.
• Or use an in-process server as it's demonstrated by MS here

If we go for the second option, it means that we have several implications:

• we must be able to inject a GrpcChannel in the ZeebeClient
• ZeebeAuthTest does not have any more sense with an in-process server and should be removed (IT covers this anyway)

@Zelldon tell me what you prefer.