Issue with Android metrics when OS kills activity about react-native-startup-time

[doomsower]

Hi! This lib is meant to be very simple. I only use it to measure cold startup time.

One way of resolving this could be to only calculate and resolve the value once from the native side. Subsequent calls would then either return the same value or reject the call (allowing conditional reporting on the JavaScript side of things).

I use it to test optimizations like code-splitting, hermes, ram bundles. So i'm Interested in in native/js combined time to first paint. This value is rough, but good enough to compare optimizations.

If you want anything more complicated, shoot a PR that allows to keep current behavior.

[john-y-pazekha]

@johanblomgren
@doomsower
Gents, please review the PR.