To answer the last question first: in a word, no. Typing and saving lag is usually the result of having too many large graphics in the text editor at once. Having graphics stored as image files in the research areas of the project (that is, everywhere outside of the Draft folder) generates very little burden. Scrivener is designed from the ground up to handle bulk amounts of material, it saves your work using a folder and file based system (like iTunes, which doesn’t slow down no matter how many mp3s you have). So it’s not at all like Word where you’ve got this ever increasingly mammoth file. Rather with Scrivener the more items you add to your binder, the more files plural you will have on your disk, encapsulated within the project. One can thus have tens of thousands of images, gigabytes of them even, without slowing the project down.
You will, most likely, notice an increase in how long it takes to back the project up while closing it, though. If you start hitting that threshold, consider the File ▸ Import ▸ Research Files as Aliases… facility, which takes the concept of linking into the binder itself, and broadens it to many different file types. It’s a good tool to familiarise yourself with, if you anticipate gigabytes in the same project where it might be a good idea to back up multiple times per day (like a typical heavily worked upon WIP).
So to return to the first question:
- Drop an image into your Research folder, just for the sake of demonstration.
- Load a “text” item from the Draft folder, place the cursor in the text editor, and use the Insert ▸ Image Linked to Document ▸ submenu to select your image.
A thumbnail will appear in the editor, but the image itself remains stored in the binder. In the text file all you get is an invisible marker that references the internal ID number of this image. The amount of text I’ve generated in the previous sentence saying that is probably more storage requirement than a link of this nature. 
The advantage here to linking to the file system is of course that “internal ID” concept being the pointer, rather than a path to a file. The latter can easily break, but with a document link you can move it around and rename it without fear of breaking the link.
As an aside, your initial reaction—of having a section of the outline that is a figure alone—is a perfectly valid and I would very useful approach to using Scrivener! The idea is to create a “text” item and insert the image into it, putting no other text within it. You can even give it a special icon to set it apart.
Where this approach becomes handy is in the ability to then refer to that figure as an individual entity, in cross-reference links, search results and so on. I will sometimes put tables, figures and other such objects in their own files specifically so I can create saved searches that build lists of figures/tables for me. It’s a more advanced approach to using Scrivener I would say, but for those working in fields where keeping track of these elements is important, it can be a good approach.
There are other approaches if the idea of breaking up your outline that much is unsatisfactory. In the past I’ve used the Comments feature to “annotated” where tables and figures are in the inspector sidebar. You can’t search for those en masse however, only one by one.