Comment

Looks like the calls to `lker.getLock` are unnecessy in `Lock` and `RLock` impelmentation

e.g. This
```golang
func (lker *Locker) Lock(key string) {
lk, ok := lker.getLock(key)
if !ok {
lk = lker.newLock(key)
}
lk.Lock()
}

func (lker *Locker) RLock(key string) {
lk, ok := lker.getLock(key)
if !ok {
lk = lker.newLock(key)
}
lk.RLock()
}
```

can be simplified to

```golang
func (lker *Locker) Lock(key string) {
lker.newLock(key).Lock();
}

func (lker *Locker) RLock(key string) {
lker.newLock(key).RLock()
}
```

Start with the Locker.Lock and Locker.RLock entry points shown in the issue, then inspect newLock to understand how lock creation and lookup work. Confirm that simplifying both methods preserves named-lock behavior, and run the repository's existing tests if available; done means both methods use the proposed direct locking flow without changing semantics.

Written by the indexing model from the issue text.