Long-term tracking issue for #151249.

TL;DR the issue is that there is less capacity in dart-internal (which engine release builds use) than flutter prod (which engine main branch builds use), which leads to longer queue times, which leads to builds which schedule and than collect sub-builds to timeout.

To prevent this in the future, we should:

• Hard-code in the top-level release builders to have longer timeouts - the repo is internal, but these are already set to timeout after 5 hours: https://dart-internal.googlesource.com/dart-internal/+/refs/heads/main/flutter-internal/engine.star#57
• edit the recipes--when scheduling sub-builds and on a release branch, to use a longer timeout https://flutter-review.googlesource.com/c/recipes/+/59360
• Increase bot capacity in dart-internal