I think David just touched on them. I will boldface each thing that is a "bad consequence".
Forgive me if I made errors in what follows.
First there are issues with the 2D polar system "embedding" into 3D.
In the obvious way, (x,y) goes to (x,y,0) in 3D.
----
In the proposal, the corresponding (r, theta) becomes (rho,phi,0) for cylindrical
coordinates.
Symbols that represent "the same thing" change:
r -> rho
theta -> phi
At least numbers in coordinates stay "natural":
(9,pi/3) -> (9,pi/3,0)
----
In the proposal, the corresponding (r, theta) becomes (r, pi/2, phi) for cylindrical
coordinates.
Symbols that represent "the same thing" change:
theta -> phi
and now there is a theta that means something completely new
And with numbers, their 2nd coordinate becomes the 3rd.
At the same time, the newly introduced coordinate is pi/2, not 0:
(9, pi/3) -> (9, pi/2, pi/3)
====
Then there are issues with converting between the coordinate systems:
Cartesian 3D <-> Cylindrical
1. x = rho cos(phi)
2. y = rho sin(phi)
3. z = z
4. rho^2 = x^2 + y^2
5. tan(phi) = y/x
6. z = z
Cartesian 3D <-> Spherical
7. x = r cos(phi) sin(theta)
8. y = r sin(phi) sin(theta)
9. z = r cos(theta)
10. r^2 = x^2 + y^2 + z^2
11. cos(theta) = z / sqrt(x^2 + y^2 + z^2)
12. tan(phi) = y/x
Cylindrical <-> Spherical
13. rho = r sin(theta)
14. phi = phi (but the cylindrical phi is coord 2, and the spherical phi is coord 3)
15. z = r cos(theta)
16. r^2 = rho^2 + z^2
17. cos(theta) = z / sqrt(rho^2 + z^2)
18. phi = phi (but the cylindrical phi is coord 2, and the spherical phi is coord 3)
Cartesian 2D <-> Polar
19. x = r cos(theta)
20. y = r sin(theta)
21. r^2 = x^2 + y^2
22. tan(theta) = y/x
It's nice that 3 and 6 are the same. And 5 and 12 are the same.
Except 5 and 12 use phis in different coordinates.
14 and 18 are weird because the phis are in different coordinates.
19--22 use letters that would make sense for Cartesian 3D <-> Spherical, but
are entirely wrong when converting Cartesian 3D <-> Spherical. The student
whose memory returns the Cartesian 2D <-> Polar equations when they needed
Cartesian 3D <-> Spherical is in trouble.
In a sense, accounting for repetition and ignoring the identities z=z and
phi=phi, there are 17 distinct equations there.
Contrast this with the "Stewart" convention, where the 4 equations from
"Cartesian 2D <-> Polar" still work perfectly as a subset of the 6 equations for
"Cartesian 3D <-> Cylindrical". In that convention, there are only 13 distinct equations.
This is a lot of "bad consequence" to weigh. And I'm back on the fence about using ISO.
It would be partly mitigated if 2D polar used phi instead of theta. But that will turn off
math teachers, where theta is in use for pre-college trig courses.
Or as has been mentioned, almost everything is nice if you use the Stewart convention
but change the meaning of phi to be "angle from the equator" instead of "angle from
the north pole". The only not nice thing about that is that it is not the ISO standard, and
it's not even a convention listed in that catalog at MathWorld. So I guess it is not used
anywhere.
I suppose making phi be "angle from the south pole" is another one to consider.