I simply looked at the Mavic specs to understand what they really mean by "2.7k".
Mavic – Specs, FAQ, Tutorials and Downloads This equates to 2704x1520 pixels, 30 frames a second.
In FCPX, I took those values and plugged it into the "New Project" information panel.
Now, is this the best mode you could use? Dunno. It won't stretch the original content you film at that resolution, so that's a plus. Almost nobody is going to view it at this resolution, since YouTube doesn't seem to list it, so that means they'll view a stretched or a downsampled version themselves. YouTube's downsample algorithm is pretty basic, meant to optimize for storage and bandwidth.
If I were shooting in 2.7k, I would probably be targeting a final project resolution of HD 1080p, the next step down. This would let FCPX downsample from the original higher-resolution video, retaining as much clarity as possible, and only treating the storage and bandwidth concerns secondarily.
Now, you mentioned that things looked worse on the screen for you -- within FCPX. Video editors often work with live temporary reduced copies (called proxy media) so that you don't suffer lag while you're working. You have to export/share a "master file" to see the best possible output FCPX can produce, rather than the working screen in your project. (Generally, I can't see much difference unless there is a LOT of finesse/finicky effects going on, besides the plain footage.)