Code review comment for lp:~raharper/curtin/trunk.fix-lp1722322

> The dirty disks mode we have just lets the existing config mode pass through.
> We can just as easily call curtin-blockmeta with an arbitrary config.
>
> It was just easiest (and immediately effective) to inherit the config from the
> install environment.

yeah, I ended up just using a different early command to lay down an MBR

>
> it'd be easy enough to make the 'Injecting early_command to dirty storage
> devices' code that is in vmtests/__init__.py accept 'cls.dirty_disks' be a
> config itself or a path to one (rather than just boolean). Then generate a
> portion of the config that would write the different config.

Yes, that's also reasonable.