Adjust URLFetcher::Create API so that object is returned as scoped_ptr.

Most interfaces were storing the object in a scoped_ptr already. This
adjusts the API so that it is a little clearer of the ownership
transfer.

A number of clients put the URLFetcher in a table and do memory
management on it themselves; this is likely templatable code
for a future CL. The scope of this CL was to change the API but
no control flow changes. Making this change found one memory leak;
(http://crbug.com/482459) has been addressed separately.

BUG=371201
TESTS=net_unittests google_api_unittests
TBR=jochen@chromium.org, thakis@chromium.org, oshima@chromium.org, armansito@chromium.org, reillyg@chromium.org, rogerta@chromium.org, stuartmorgan@chromium.org, wez@chromium.org, pavely@chromium.org, rouslan@chromium.org

Review URL: https://codereview.chromium.org/1117703002

Cr-Commit-Position: refs/heads/master@{#327901}

1 files changed, 1 insertions, 4 deletions

diff --git a/components/captive_portal/captive_portal_detector.cc b/components/captive_portal/captive_portal_detector.cc
index b7b8668..ceeffe6 100644
--- a/components/captive_portal/captive_portal_detector.cc
+++ b/components/captive_portal/captive_portal_detector.cc
@@ -34,10 +34,7 @@ void CaptivePortalDetector::DetectCaptivePortal(
   detection_callback_ = detection_callback;
 
   // The first 0 means this can use a TestURLFetcherFactory in unit tests.
-  url_fetcher_.reset(net::URLFetcher::Create(0,
-                                             url,
-                                             net::URLFetcher::GET,
-                                             this));
+  url_fetcher_ = net::URLFetcher::Create(0, url, net::URLFetcher::GET, this);
   url_fetcher_->SetAutomaticallyRetryOn5xx(false);
   url_fetcher_->SetRequestContext(request_context_.get());