Fix duplicate key test case

[DifferentialOrange]

Issue is described in #105. After further investigation, it was found that it was meant to be "insert structure using space name" test case that must process with success instead of "duplicate key error". Test case and corresponding output was changed to be consistent in first commit.
Before patch test case:

go-tarantool/example_test.go

Lines 121 to 125 in 61f3a41

	// insert new tuple { 11, 1 }
	resp, err = client.Insert("test", &Tuple{Id: 10, Msg: "test", Name: "one"})
	fmt.Println("Insert Error", err)
	fmt.Println("Insert Code", resp.Code)
	fmt.Println("Insert Data", resp.Data)

Since there is a duplicate key error test case in another file, coverage has not decreased.

go-tarantool/tarantool_test.go

Lines 437 to 443 in 61f3a41

	resp, err = conn.Insert(spaceNo, &Tuple{Id: 1, Msg: "hello", Name: "world"})
	if tntErr, ok := err.(Error); !ok || tntErr.Code != ErrTupleFound {
	t.Errorf("Expected ErrTupleFound but got: %v", err)
	}
	if len(resp.Data) != 0 {
	t.Errorf("Response Body len != 1")
	}

Closes #105

[Totktonada]

Seems meaningful for me (but I don't know this code in fact).

[LeonidVas]

Hi! Thank you for the patch.
See several commits below:

• I think that "Make test case consistent with comments" is more a workaround rather than a fix. Because we have lost "Duplicate key ..." test case. But it seems ok to me. In fact, using an error message to check an error is bad practice, an error code should be used for this purpose.
• Add a description of a problem to a commit message (something like this in the issue description). This allows to understand the reason of changes in the future.

[DifferentialOrange]

* I think that "Make test case consistent with comments" is more a workaround rather than a fix. Because we have lost "Duplicate key ..." test case. But it seems ok to me. In fact, using an error message to check an error is bad practice, an error code should be used for this purpose.

It may seems like this, so at first I wanted to add a separate test case for Duplicate error. But such test case already is covered by tarantool_test.lua with error code assert. See more in PR description.

[LeonidVas]

* I think that "Make test case consistent with comments" is more a workaround rather than a fix. Because we have lost "Duplicate key ..." test case. But it seems ok to me. In fact, using an error message to check an error is bad practice, an error code should be used for this purpose.

It may seems like this, so at first I wanted to add a separate test case for Duplicate error. But such test case already is covered by tarantool_test.lua with error code assert. See more in PR description.

Ouch! I missed it. OK.

[DifferentialOrange]

* Add a description of a problem to a commit message (something like this in the issue description). This allows to understand the reason of changes in the future.

Added a description and duplicated PR commets to commits.

[Totktonada]

@funny-falcon Aren't you mind to glance on these changes?