I mean, if your "design time" package is just going to reuse all the units from your "run time" package, but the .PAS files for both are all in the same folder to begin with, then just put everything in one package!
NO, that is a terrible idea! Do you actually understand the difference between a
runtime package and a
design-time package?
Lets take something I know -
OnGuard (security components). Originally everything was lumped into one Lazarus design-time package. The problem - I couldn't use it with fpGUI, because the design time package references Lazarus IDE and LCL units, and I was coding a pure fpGUI only application. Technically the OnGuard components are totally GUI Toolkit independent, but because of using the design-time package, it forced a LCL dependency.
So what I did was to split the OnGuard into two packages. A runtime package which is 100% non-GUI. This means anybody can use the runtime package with fpGUI or even Console or WebServer applications. Then I created a design-time package, and added the runtime package as a required dependency, and added a unit that contains the code that allows it to register on the Lazarus IDE's component palette (something I never require in my projects). So now any type of project (LCL, fpGUI, Console, Web etc) can use OnGuard components.
This problem is moot if you don't use Lazarus Packages, because then you would simply set the Unit Search Path for your project and instantiated any classes you want via code. Though I have seen badly designed components that have artificial (unnecessary GUI) dependencies, which need to be refactored first.