You could use Hitfilm to remove the fisheye, then export that flatter video, reload it and use that in Mocha Hitfilm to do the track on instead.
Or, if you wanted to retain the fisheye look, do the above to flatten it, then put the resulting camera and your text in a separate composite shot on its own with no video.
Put that composite inside another one with the original fisheye video footage, then add a reverse of the fisheye removal to the text composite to bend it correctly back onto the original video.
Or use Two-Point tracking in Hitfilm, which will give you flat text, but position, scale and rotation on your text to make it look mostly correct. It depends if you want it perfect, or close enough, but at least not wobbling about.