Luke:
Do note that with 2.1.0E, any XML file in your sandbox that is not associated with a web form will prompt for a web form association when you go to edit it. That way, if you are importing valid XML, use can select which of the web forms associated with your web projects you wish to associated it with for purposes of editing. We hope to make this intelligent, so that only that web form that is suitable for that XML content type will be presented to the user (and then, there only selection is really to use a web form for editing or not).
For bulk migrations, I would actually recommend using a special "load" sandbox instead. The reason for this is that you can collect a number of changes and then bulk process using a scheduled action. In a normal space, a rule will execute on each item. If you are bulk loading a lot of data, it may very well be much more efficient to just process them in bulk. Similarly, because you only have to configure that workflow once, regardless of what directories and subdirectories you may be creating to ingest the content, it will be simpler to import into a sandbox via CIFS than to use a normal space, which will require you to set a rule on each subspace you create for foldering.
Now, if you are going to be using spaces and promoting to a web project, and if for some reason you do not want to use our pre-built action for copying an item to a space, make sure you use the interstore copy service in our JavaScript API. This is responsible for moving assets from a space to a sandbox. It also handles auto-creation of directories to replicate whatever space structure you have, and also copies over all associated metadata as well.
Kevin