On FATAL, log which unsupported encoder the caller wanted us to create

On FATAL, log which unsupported encoder the caller wanted us to create

Hopefully, this will make it easier to figure out what's wrong the
next time this happens.

BUG=526478
Committed: https://crrev.com/6aae75728a68bf8e59f75e3a87bc8aefb2ed3907
Cr-Commit-Position: refs/heads/master@{#9844}

[kwiberg-webrtc, 2015-09-02 09:41:28]

kwiberg@webrtc.org changed reviewers:
+ henrik.lundin@webrtc.org

[kwiberg-webrtc, 2015-09-02 09:42:46]

♫

[hlundin-webrtc, 2015-09-02 10:37:25]

lgtm

[kwiberg-webrtc, 2015-09-02 11:14:22]

The CQ bit was checked by kwiberg@webrtc.org

[commit-bot: I haz the power, 2015-09-02 11:14:31]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1313073008/1
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1313073008/1

[commit-bot: I haz the power, 2015-09-02 12:05:05]

Committed patchset #1 (id:1)

[commit-bot: I haz the power, 2015-09-02 12:05:15]

Patchset 1 (id:??) landed as
https://crrev.com/6aae75728a68bf8e59f75e3a87bc8aefb2ed3907
Cr-Commit-Position: refs/heads/master@{#9844}

[tommi, 2015-09-02 12:55:12]

tommi@webrtc.org changed reviewers:
+ tommi@webrtc.org

[tommi, 2015-09-02 12:55:12]

Is it still right to crash in this case?  If we log more information and crash,
that log isn't actually going anywhere.

[kwiberg-webrtc, 2015-09-02 13:13:34]

On 2015/09/02 12:55:12, tommi (webrtc) wrote:
> Is it still right to crash in this case?

I'd argue yes. This far down, it's the caller's fault for passing us bad
arguments. The bug should be fixed by making higher layers not do that. But to
do that, we need to figure out what sort of mistake they did.

> If we log more information and crash,
> that log isn't actually going anywhere.

The reporter saw the message:
https://code.google.com/p/chromium/issues/detail?id=526478#c5
Which at the time didn't print anything interesting.

But you're right for any users not savvy enough to run from a terminal. Could
another sort of logging call work around the problem?