Revert removal of `--no-open` flag to `ionic cap run`.

The capacitor 3 support added to the CLI changed the behavior of `ionic cap run`. It now does one of two things to "launch" the native project:
- Deploy it directly to a target emulator (either with the `--target` flag or interactively)
- Run `cap open` to open the native project

However, I would like it to do neither of these things. My previous workflow used the `--no-open` flag to use `ionic cap run` to simply manage the angular build server and the syncing of the `capacitor.config.json`, and then I manually executed the native part of the build process (which is necessary for my project, as it has many different schemes/flavors to build). I would like to request that this workflow continue to be supported. The currently available `--open` flag is less than ideal, because it will spawn a new Android Studio process every time it's executed.

Thanks for all your hard work on this project!