I envy you the fact that it is only a word count that counts. I am in a system where everything has to fit a certain number of pages with 12pt Times-Roman and 2.5 cm margins.
With the word limit as your main limitation you have the possibility to illustrate your application more freely. If suitable, I strongly suggest you try to come up with conceptual graphics to strengthen the text and make difficult concepts easier to grasp. This can be a huge advantage. Since your are not limited to pages you can afford complex and larger illustrations than otherwise possible.
As for typesetting, you use LaTeX (as do I) which produces excellent text. Trust the LaTeX settings for text width, line-spacing etc. to get a readable text. Too long or too short lines makes the text difficult to read. See the geometry package for details if you are not already using it. EDIT: I do not think two-column format is good for proposals. It is a way to put as much text into a small area and also involves using smalller type sizes, typically around 9 in order to get enough characters into a line to make it readable. Since the number of pages is not a limit, I would aim for something which is as easily read as possible, typically 11-12 pt single columns. the margin widths can be determined so as to get reasonable line lengths. I do not think wide or narrow margins make a difference but wider margins make the page more attractive than narrow (basically a typographical design issue).
EDIT: With one-column format you can add figures at full width and with the no-page-limit do not have to worry about the size of the graphics (in the sense of having to make it as small as possible so to not use up space. I think a straight-forward simple formatting is best. There is no need to be overly creative. What conveys the message without resistance is the best.
One of the more useful concepts in grant writing I have come across is given in the book Writing Successful Science Proposals by Andrew J. Friedland and Carol L. Folt which outlines a two-paragraph summary, akin to an abstract whith which you introduce your proposal. The point of this is to quickly and concisely convey the key points of the proposal to the reader. This involves starting from the big picture metion key gaps in knowledge, how to tackle them (methods), preliminary results (if existent) and expected results. I obviously cannot copy the book content here but the point is to make the reader of the proposal completely clear over what to expect in the proposal and so all details given is just putting substance to the known structure. I believe this is what you aim for in what you call an overview section.
EDIT: After having started your text with a brief introduction, you simply follow up by the usual type of structure:
- background (to focus on the gap of knowledge to be targeted)
- project outline
- time table (milestones; can be a gantt plot)
- if applicable: research group/resources
For references I would go for footnote type references since the harvard-style citations uses up many characters. It is also possible to abbreviate the reference list by omitting the paper title, similar to what is done in Nature and Science.