Remove a pointless DCHECK and fix a comment.

https://crrev.com/c/2426086 introduced a pointless DCHECK and also had a
comment on ShouldCreateNewHostForCrashedFrame that was left over from
the approach of skipping early commit on all non-live frames.

Bug: 1072817
Change-Id: I2a8c4ad6147a6941baf6d3fdd7358a80c74e41fd
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2469346Reviewed-by: Alex Moshchuk <alexmos@chromium.org>
Commit-Queue: Fergal Daly <fergal@chromium.org>
Cr-Commit-Position: refs/heads/master@{#820683}

content/browser/renderer_host/render_frame_host_manager.cc

@@ -829,8 +829,6 @@ RenderFrameHostImpl* RenderFrameHostManager::GetFrameHostForNavigation(
 
     if (!render_frame_host_->IsRenderFrameLive() &&
         !recovering_without_early_commit) {
-      // We should only do this if the current RFH is not live.
-      DCHECK(!render_frame_host_->IsRenderFrameLive());
       // The current RFH is not live. There's no reason to sit around with a
       // sad tab or a newly created RFH while we wait for the navigation to
       // complete. Just switch to the speculative RFH now and go back to

content/public/common/navigation_policy.h

@@ -93,8 +93,8 @@ struct CONTENT_EXPORT NavigationDownloadPolicy {
 CONTENT_EXPORT bool ShouldCreateNewHostForCrashedFrame();
 
 // If this is false we continue the old behaviour of doing an early call to
-// RenderFrameHostManager::CommitPending when we are replacing an inactive
-// frame (crashed or just newly created but never live).
+// RenderFrameHostManager::CommitPending when we are replacing a crashed
+// frame.
 // TODO(https://crbug.com/1072817): Stop allowing this.
 CONTENT_EXPORT bool ShouldSkipEarlyCommitPendingForCrashedFrame();