If you introduce a new rail type into your rail network you can't use your existing fleet of trains on that section reducing the ROI on that train engine or carriage. Also, any train you purchase for the new rail type will only ever work on that system lowering their profitability in the long term.
A million times this. Mag-lev only works for either super dense routes where the added cost as you describe can be displaced by the immense value add of shorter and generally more comfortable travel. Or in nations that can force through decisions from the top down, such that cost becomes almost a non-factor like China. Rail in general across the western world is a weird mix of nationalized and privately owned companies and operators, such that introducing mag-lev with the intent to replace conventional rail would require compensation to the private companies who have invested billions in the current infrastructure else they simply won't be part of the new one, with all the issues that entail.
From an environmental standpoint it's also really hard to see an ROI in scrapping something that works in favor of mining, constructing and spending intense amounts of energy in all forms to build something better but only moderately so. The biggest improvement is moving from trucks to (electric) train for freight, going from electric train to mag-lev is only slightly better so the ROI just won't be there.
I would suspect it would be complex to design mag-lev for all the various types of loads trains for be subject to. Wheels are fairly versatile and have a wide range of loads.
Also the fact that 'less moving parts' doesn't mean lower complexity or maintenance cost. Train wheels are a very robust and efficienct mechanism and most train designs are not being limited by them.
Very robust because they have 300 years of research, innovation, materials science and manufacturing in them. Making them incredibly stellar, well understood, damn near perfect technology for what they do.
There are some concepts for hybrid maglev-rail tracks that would at least solve the first point, similar to how rail was electrified over time. It would still be very expensive though.
maglevs need classical wheel systems anyway because there might be a power outage, so simply having wheels that are compatible with the local rail system is a brilliant idea.
add in a tiny propulsion system so they can use the normal tracks at low speed without the help of the maglev tracks and you can sort of blend the two systems together in critical locations like switches and train stations.