Add support for better "try-lock" pattern for Locker<Lock>

[Chris Dumez]

Add support for better "try-lock" pattern for Locker<Lock>. Maybe:
```
Locker locker { DeferLock, m_lock };
if (!locker.tryLock())
    return;

// Do something.
```

I think it would be nicer than the current:
```
if (!m_lock.tryLock())
    return;

Locker locker { AdoptLock, m_lock };
// Do something.
```

[Chris Dumez]

Created attachment 429536 [details]
WIP Patch

Very minimal patch for now as EWS didn't like it last time I tried this approach.

[Chris Dumez]

Created attachment 429537 [details]
WIP Patch

[Chris Dumez]

Created attachment 429540 [details]
WIP Patch (more complete)

[Chris Dumez]

Created attachment 429542 [details]
WIP Patch (more complete)

[Chris Dumez]

Comment on attachment 429542 [details]
WIP Patch (more complete)

View in context: https://bugs.webkit.org/attachment.cgi?id=429542&action=review

> Source/WebCore/Modules/speech/SpeechRecognitionCaptureSourceImpl.cpp:88
> +    Locker locker { DeferLock, m_dataSourceLock };

./Modules/speech/SpeechRecognitionCaptureSourceImpl.cpp:109:5: error: writing variable 'm_dataSource' requires holding mutex 'm_dataSourceLock' exclusively [-Werror,-Wthread-safety-analysis]

EWS gets this error. I don't see what I am doing wrong here and this works fine locally :/

[Chris Dumez]

Add https://bugs.webkit.org/attachment.cgi?id=429537&action=prettypatch was building fine so the pattern seems to be working, at least in some cases..

[Radar WebKit Bug Importer]

<rdar://problem/78689398>