net/http: fix ineffective break in etag scanning

In particular, this lead to the code accepting invalid ETags as long as
they finished with a '"'.

Also remove a duplicate test case.

Change-Id: Id59db3ebc4e4969562f891faef29111e77ee0e65
Reviewed-on: https://go-review.googlesource.com/39690
Run-TryBot: Daniel Martí <mvdan@mvdan.cc>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

src/net/http/fs.go

@@ -315,7 +315,7 @@ func scanETag(s string) (etag string, remain string) {
 		case c == '"':
 			return string(s[:i+1]), s[i+1:]
 		default:
-			break
+			return "", ""
 		}
 	}
 	return "", ""

src/net/http/fs_test.go

@@ -1264,10 +1264,10 @@ func Test_scanETag(t *testing.T) {
 		{`"etag-2"`, `"etag-2"`, ""},
 		{`"etag-1", "etag-2"`, `"etag-1"`, `, "etag-2"`},
 		{"", "", ""},
-		{"", "", ""},
 		{"W/", "", ""},
 		{`W/"truc`, "", ""},
 		{`w/"case-sensitive"`, "", ""},
+		{`"spaced etag"`, "", ""},
 	}
 	for _, test := range tests {
 		etag, remain := ExportScanETag(test.in)