Not sure where "megadensity" comes from. The correct term would be resolving power. The megapixel rating is an indication of the level of detail the lens can resolve from a standard resolution test chart. Megapixel when used in the context of the lens, is not really a true term in so far as it's not scientific.
Basically though, the higher the resolving power, the potentially sharper the image. Personally, I think you can never go too sharp. The worst case scenario is the lens can resolve more detail than the sensor can record - the result is the picture will be maxx'ed out for sharpness and detail, with some headroom in the lens to spare. However, get too a low resolving power, and the image won't be as sharp or detailed as it could be. The only downside to buying too much resolving power is cost, but provided lenses are a similar price, as I see it, you have very little to lose going for the higher resolving power.
Note as well here, lenses aren't just about resolution alone. Low distortion often through the use of an aplantic lens, can be just as important as most cheaper lenses are not as sharp all the way across the lens and have distortion towards the edges that can affect the picture quality for non central objects.
If looking for lenses, have a look at the Backbone website:
https://www.back-bone.ca/ There maybe a lens there you could potentially use. Again I'd say concentrate on low distortion as well as resolving power.
Also, one final factor, not talked about on here usually is the relationship between the number of sensor pixels and identifiability of faces / objects. To this end the higher pixel count sensor is king. In CCTV circles, as a rule of thumb the aim is to achieve 80-100 horizontal pixels across a face so as to achieve recognisability. The relationship this has to cameras is:
1. The wider the lens, the more area you are spreading the sensor's pixels out over, so the lower the pixel density
2. The higher the sensor rating in pixels, the more pixels your likely to have in any given area eg over a face or number plate increasing readability
3. The distance from the lens also plays a part because as you enlarge the image to make it recognisable, you effectively reduce the pixel density.
That's my understanding at least. A little simple guide here from a CCTV manufacturer:
https://www.axis.com/files/feature_articles/ar_perfect_pixel_count_55971_en_1402_lo.pdf
Must admit, to the end that sensor pixel count has an influence on picture recognition quality, I am a little surprised now that cheap 4K CCTV cameras are out there, that we haven't seen 4K sensors finding their way into dashcams. Maybe the sensors are just still a little too expensive for this purpose.
However, I believe there are potentially gains to be made from a higher quality lens.