What you're seeing is an effect similar to "Posterization" (https://en.wikipedia.org/wiki/Posterization) -- though that is usually considered purposeful, and here the similar consequences are unintended.
Essentially, that "banding" (you called it a "wave" effect) of the colors around the lights is actually present in both images. You can see it if you zoom in on your top image. It's an inherent property of the undersampling -- color or chroma undersampling -- of the image sensor of your camera.
which leads to aliasing artifacts: https://en.wikipedia.org/wiki/Aliasing
Your camera, depending on its color model and/or color profile, has only a finite (read: bit limited) number of color sample values that it can represent, but the light (the electromagnetic radiation eminating or reflecting from objects/scenes) is essentially continuous. The light also, from a frequency domain view, contains nearly all frequencies; obviously it has a stronger energy related to some (or a few) frequencies more than most, but in terms of determining the fundamental frequency of the light you're trying to capture, it's a value that's well beyond the limits of your camera, and thus double that value (the theoretical Nyquist frequency required for non-aliased sampling) is beyond the bit-depth limitations of your camera's color profile for sure.
Remember that color is a semantic labelling of the wavelength of electromagnetic radiation and the wavelength of a signal is the inverse of its frequency, thus color is also a semantic labelling of frequency. So if your camera is failing to accurately record (read: sample) the frequency content of the scene light that you're trying to capture, you will get color aliasing effects.
That's the "physics" explanation for "why?".
The practical question is "Why does changing to CMYK exacerbate this posterization/aliasing effect around the lightpoles?".
That is because you're asking the software to convert, transform, and/or resample your color data to a new color profile, CMYK. Again, every color profile is bit-limited due to our finite digital technology, so each color profile can only represent a limited number of wavelengths (read: frequencies, i.e., "colors").
To you, the first picture looks "better" because the color profile for that picture has more bit-depth (better sampling) in the frequency range associated with the colors of light around the lightpoles, thus their "glow" appears nice and smooth as it transitions into the darkness of the night sky. When you use Photoshop -- which could be set for printing, thus defaulting to CMYK -- or set Lightroom to CMYK, your profile now has less bit-depth associated with those colors (frequencies) and so the undersampling creates aliases (high-frequency artifacts) which exacerbate the transitions from smooth to "rough", making the posterization effect.
Aliasing can be mitigated in pre-processing or post-processing with Low-Pass Filtering (blurring or smoothing), but this also reduces spatial details.
Unless your color profile of your camera and your display software and your computer monitor can all meet the theoretical Nyquist Frequency for the frequencies of the colors in the scene you want to sample, you will always see some variation of this effect.
That's why, also, when you switch applications, switch computer monitors, or switch cameras (or switch any combination of those) when capturing images of the same scene, you're going to see variations of this effect as well.