dashboard: double darwin-amd64-longtest timeout scale

Using the same timeout scale value of 5 for darwin-amd64-longtest as
for all other -longtest builders has proven not to work well. Scale it
by another factor of 2 to account for its overhead due to emulation.

The few data points I've manually collected so far suggest this is
enough to allow it to pass reliably, and we will be able to collect
more as part of release testing.

For golang/go#29252.
Fixes golang/go#60919.

Change-Id: Id740248703903eb1b351d7bf284f8749ab0386d0
Reviewed-on: https://go-review.googlesource.com/c/build/+/506515
Run-TryBot: Dmitri Shuralyov <[email protected]>
Reviewed-by: Dmitri Shuralyov <[email protected]>
Reviewed-by: Michael Pratt <[email protected]>
Reviewed-by: Bryan Mills <[email protected]>
Auto-Submit: Dmitri Shuralyov <[email protected]>
TryBot-Result: Gopher Robot <[email protected]>
Reviewed-by: Carlos Amedee <[email protected]>

Author: dmitshur

dashboard/builders.go

@@ -2672,7 +2672,11 @@ func init() {
 			return b
 		},
 		env: []string{
-			"GO_TEST_TIMEOUT_SCALE=5", // give them lots of time
+			// We use a timeout scale value of 5 for most longtest builders
+			// to give them lots of time. This particular builder is not as fast
+			// as the rest, so we give it 2x headroom for a scale value of 10.
+			// See go.dev/issue/60919.
+			"GO_TEST_TIMEOUT_SCALE=10",
 		},
 	})
 	addBuilder(BuildConfig{
@@ -2696,7 +2700,7 @@ func init() {
 			// Increase the timeout scale for this builder: it was observed to be
 			// timing out frequently in
 			// https://go.dev/issue/55311#issuecomment-1571986012.
-			///
+			//
 			// TODO(bcmills): The darwin-amd64-longtest builder was running extremely
 			// slowly because it was hitting swap. Race-enabled builds are also
 			// memory-hungry — is it possible that the -race builder is also swapping?