Use --lto-O0 for optimizeForSize targets

This is another place where we can trade performance for additional
binary size savings.

Test: presubmit
Change-Id: Icf9600ef2ca4092e4adccc0c26465b542b038f7b
diff --git a/cc/lto.go b/cc/lto.go
index 4444152..f3af7d2 100644
--- a/cc/lto.go
+++ b/cc/lto.go
@@ -110,7 +110,7 @@
 		var ltoLdFlags []string
 
 		// Do not perform costly LTO optimizations for Eng builds.
-		if Bool(lto.Properties.Lto_O0) || ctx.Config().Eng() {
+		if Bool(lto.Properties.Lto_O0) || ctx.optimizeForSize() || ctx.Config().Eng() {
 			ltoLdFlags = append(ltoLdFlags, "-Wl,--lto-O0")
 		}