Well indeed such coordinate systems exist but you've still got the cost of translating a large body of cartography into them.
Plus as pointed out by another commentor the number of KMs around the earth is not a round number, so you need to do some fudging to get a metricised spherical coordinate system (and you want a spherical coordinate system because the 60th of a degree = 1nm equivilance is very useful for simplfying the mathematics required for navigation).
The cost of translating is a mouse click [after a software pro writes the code behind the control].
Modern, computerized charts don't need to lie flat or use only one projection visualization. Furthermore, you can add up-to-date ocean current, jet stream, wind, wave, and weather information to a computerized chart, along with a vehicle performance model, and automatically optimize a route for fastest travel time or lowest fuel consumption.
Nowadays, the unit of distance on the printouts and reports is only needed for human readability and loss-of-power emergencies. Once you have planned out your route, you can scale your backup hard copy with whatever unit your navigator feels most comfortable with, but inside the computer, the software most definitely has a "native" measurement unit, and those are usually in the base SI unit. All other units are converted from that native representation as necessary.
You can navigate your vessel in any coordinate system you like, as long as the conversion functions are coded correctly. And once they are, all charts are translated into all coordinate systems. If you like constant-bearing lines on Mercator projections, just print it out that way. If you like following currents and great circles, just make sure your navigation computer has a backup generator.
The fudging you mention is 99.9% handled just by using the WGS84 ellipsoid. If you're doing any fancy rocket science, you also need a somewhat detailed gravitational model mapped on top of it. Those sorts of calculations are most easily done in an ECR cartesian coordinate system, which easily translates to ECEF cartesian, which somewhat less easily translates to geodetic coordinates--it is actually easier to translate that in the other direction. Unfortunately for that math, no one knows where their house is on the Earth in XYZ coordinates. The amount of error creeping into your double-precision floats is almost insignificant at the scale you need with one or two extra conversions.
The math only needs one arbitrarily chosen unit. I have never seen nautical miles chosen as the unit for length.
Plus as pointed out by another commentor the number of KMs around the earth is not a round number, so you need to do some fudging to get a metricised spherical coordinate system (and you want a spherical coordinate system because the 60th of a degree = 1nm equivilance is very useful for simplfying the mathematics required for navigation).