Add RTCConfiguration getter and setter methods.

webrtc/api/peerconnectioninterface.h

Index: webrtc/api/peerconnectioninterface.h
diff --git a/webrtc/api/peerconnectioninterface.h b/webrtc/api/peerconnectioninterface.h
index 08a131920e3be649746ebbc083cdf8b4f64d7a69..b8984cd6ed8d560f4bb0eac1892335fe2bef3564 100644
--- a/webrtc/api/peerconnectioninterface.h
+++ b/webrtc/api/peerconnectioninterface.h
@@ -223,6 +223,32 @@ class PeerConnectionInterface : public rtc::RefCountInterface {
 
   // TODO(hbos): Change into class with private data and public getters.
   struct RTCConfiguration {
+    // Getter and setter methods for the flags which are used by
+    // applications, chrome in particular.
+    bool get_enable_dscp() { return enable_dscp.value_or(false); }
+    void set_enable_dscp(bool flag) { enable_dscp = rtc::Optional<bool>(flag); }
+
+    bool get_enable_cpu_overuse_detection() {
+      return cpu_overuse_detection.value_or(true);

[perkj_webrtc, 2016/03/29 15:25:11]

Use in this cl? ie- Stop using the to be private structs.

Name cpu_overuse()
set_cpu_overuse(bool enabled)
?

......

[nisse-webrtc, 2016/03/30 08:50:24]

(Some important words seem to be missing). My intention is to land this, update
chrome, and then rebase and land https://codereview.webrtc.org/1818033002/. At
that point, webrtc won't really do any read accesses to these individual flags
in RTCConfiguration, just pass on a pointer/reference to the MediaConfig where
they are stored.

And for write accesses, it's the (deprecated) code converting constraints to
config values, which will be using FindConstraint or (before my changes)
ConstraintToOptionalBool. Going via the setter function there is just making
things unnecessarily complicated.

But there are a few uses in the test cases, and there it makes sense to update
to the getter and setter methods.
I think just cpu_overuse is confusing. "cpu_adaptation", maybe?

I would also like (in a later cl) to make the name of the MediaConfig fields a
bit more consistent. There, we have an enable_ prefix on two flags, disable_ on
one flag, and one flag with no prefix. So really inconsistent. Is it reasonable
to have "enable_" prefixes on the variables, but not on the methods?

+    }
+    void set_enable_cpu_overuse_detection(bool flag) {
+      cpu_overuse_detection = rtc::Optional<bool>(flag);
+    }
+
+    bool get_enable_suspend_below_min_bitrate() {
+      return suspend_below_min_bitrate.value_or(false);

[perkj_webrtc, 2016/03/29 15:25:12]

dito

+    }
+    void set_enable_suspend_below_min_bitrate(bool flag) {
+      suspend_below_min_bitrate = rtc::Optional<bool>(flag);
+    }
+
+    bool get_enable_prerenderer_smoothing(bool flag) {

[perkj_webrtc, 2016/03/29 15:25:11]

|flag| is unused

+      return !disable_prerenderer_smoothing;
+    }
+    void set_enable_prerenderer_smoothing(bool flag) {
+      disable_prerenderer_smoothing = !flag;
+    }
+
     static const int kUndefined = -1;
     // Default maximum number of packets in the audio jitter buffer.
     static const int kAudioJitterBufferMaxPackets = 50;