Use pointer receiver for type-defined errors

[wolfogre]

The old code use value receiver for type-defined errors like ErrTaken, it makes ErrTaken and *ErrTaken are both valid errors.

When I saw the the value receiver, I thought the way to check errrors should be

var target redsync.ErrTaken
if errors.As(err, &target) {
	// ...
}

Unfortunately, it didn't work. Since the code returned &ErrTaken.

redsync/mutex.go

Line 327 in b292c9f

err = multierror.Append(err, &RedisError{Node: r.node, Err: r.err})

---

And I also see mix-use of the two types in the repo, the test code uses RedisError as an error:

redsync/error_test.go

Lines 19 to 22 in b292c9f

	err := RedisError{
	Node: 0,
	Err: v,
	}

But here it uses *RedisError as an error:

redsync/mutex.go

Line 327 in b292c9f

err = multierror.Append(err, &RedisError{Node: r.node, Err: r.err})

---

So, I believe it's be a good idea to use pointer receiver for type-defined errors, just like most errors in golang standard packages like url.Error and os.PathError. It doesn't break anything, since the right ways to handle such errors are the same before or after this PR:

var target *redsync.ErrTaken
if errors.As(err, &target) {
	// ...
}