Change error message in case of successful license application

[all-seeing-code]

Before this code change in case of successful license application, http response would look like:
{ "errors": [ { "message": "Done", "extensions": { "code": "Success" } } ] } .
This is confusing since although the license has been correctly applied, it still shows the success in the error field. This PR changes the above response to:
{ "code": "Success", "message": "License applied." } to remove any confusion.

Fixes #GitHubIssue 4965
Fixes Dgraph-1135

---

This change is 

Docs Preview:

[jarifibrahim]

Reviewable status: 0 of 1 files reviewed, 1 unresolved discussion (waiting on @anurags92, @manishrjain, and @parasssh)

---

dgraph/cmd/zero/license_ee.go, line 133 at r1 (raw file):

		return
	}
	x.Check2(w.Write([]byte(`{"code": "Success", "message": "License applied."}`)))

w.Write call could return an error. If it fails, we will end up killing the server because Check2 contains log.Fatal. We should handle the error appropriately here.

[parasssh]

Reviewable status: 0 of 1 files reviewed, 2 unresolved discussions (waiting on @anurags92, @manishrjain, and @parasssh)

a discussion (no related file):

---

[parasssh]

Reviewable status: 0 of 1 files reviewed, 2 unresolved discussions (waiting on @anurags92, @manishrjain, and @parasssh)

a discussion (no related file):

---

[martinmr]

Reviewed 1 of 1 files at r1.
Reviewable status: all files reviewed, 2 unresolved discussions (waiting on @jarifibrahim, @manishrjain, and @parasssh)

a discussion (no related file):

Previously, parasssh wrote…

:lgtm:

once the point Ibrahim raised is fixed. I added a bit of context on why Check2 is used sometimes in the codebase.

---

---

dgraph/cmd/zero/license_ee.go, line 133 at r1 (raw file):

Previously, jarifibrahim (Ibrahim Jarif) wrote…

w.Write call could return an error. If it fails, we will end up killing the server because Check2 contains log.Fatal. We should handle the error appropriately here.

Agree. Check2 is used sometimes but that's only to write to a string buffers because that's very unlikely to return an error when writing small strings.

But we don't have the same kind of control on the w writer. An error here should probably print a warning in the server logs at the least.