As much as I’m in favour of 3000K LED street lighting, I find the colour of 2700K and 2200K to be even more pleasant, though I have only experienced 2700K once whilst I’ve never experienced 2200K. It really makes me wonder, why are they being used less despite having a more pleasant colour and are generally less harsh than 4000K and colour temperature above?

I am convinced that 9 times out of 10, it's ignorance.

My city is a complete hodgepodge of colors, though the replacements of the PL and TL street lights in residential areas, do appear to 2700k on the residential streets, and 3000 or 4000 on the roads that exit said streets.
Cycle paths through parks also get 2700k which actually look really nice.

The province, however, has installed - i think - 6500k on the highways and provincial main roads and i hate them. They are extremely bright, way brighter than needed, and the extreme color temperature means that they're very bad for skyglow and nature.
Provincial minor rural roads seem to get 3000k which is alright. My own residential street also appears to have gotten 3000k, though it's hard to estimate color temperature by eye. The main road still has SON-Ts.

There is a large 3/4 lane roundabout that always was lit by SOX, which now has 2700k very warm white LEDs which look neat, but the roads that feed it, are all different colors. I even know an installation that has on 1 double-arm pole, one 3000k and one 2200k LED, it's hilarious.

I think ignorance is the biggest factor here. People who do take notice and care about the correct CCT of light sources are in the minority. I know some people who do not notice when a 2-lamp fluorescent fixture has one 3000K lamp and one 6500K lamp. Go figure...

One thing that is a real issue in the Netherlands, is because people have been voting for small government type parties, that a LOT of technically schooled personnel from municipalities and provinces, has been fired to reduce spending on government employee salaries. Whether a municipality still has an engineering team, very much depends on what parties the citizens vote for.

This has lead to the problem that those who make the decisions to re-do the entire lighting system of a city, sometimes do not know ANYTHING about street lighting. They are fully reliant on the expertise of external contractors.

Every contractor has their own philosophy about what they think is best. So for each new project, and each new tender that has to be offered to all countries within the EU, different people design the lighting system for that project. And that of course leads to a different appearance and functionality. 

Obviously, this does not really save money at all. The idea of not having an engineering staff at every municipality is nice, but the reality is that it results in spending just as much, but now with the added downside of being fully dependent on other parties and still ending up with a very unstructured lighting philosophy for a town.

My city has chosen for complete replacement of fixtures. Amsterdam has a LOT of Philips Iridium fixtures, and i have the feeling that a considerable part of those have been refurbished with LED panels rather than complete replacement, but i'm not sure - but i rarely am in Amsterdam at night time these days.

Finally, there is one objectively good reason for choosing a higher color temperature - the human eye has higher resolution/accuity at higher color temperatures. The old SOX lamps were not great in that regard.

Cri is important, 4000k led beats son (2200k) any day IMHO. I find it kind of shitty driving under son streetlights because you can't see for crap. Nice in collections and the odd missout but not for all lighting. 2700k and 3000k seems fine though.

Maybe the phosphors for leds are more expensive? Usually warm white led bulbs cost more than daylight or coolwhite equivalents here.

No, that's not a significant cost factor in the grand scheme of things. Better CRI costs more, otherwise cool and warm white are pretty much identical in cost.

It all comes down to cost and efficiency.

LEDs with warm color temperatures are nowhere near as efficient (in terms of both lumen output and lumen efficacy) as those with cool color temperatures. Unless there are special mandates for street lights to be warm or amber, that is the primary reason why LED luminaires with a CCT of 3,500K or greater are so common.

Yttrium aluminum garnet (YAG) is typically used for cooler CCTs; a blend of YAG and a silicon-based nitride (YG630). The blend will obviously be more expensive to use.

Regardless of the phosphor cost, the lumen output so the efficacy reduction is the biggest cost.
The problem is in bare physics:
From a single blue photon the phosphor can not make more than a single photon anywhere in the visible range. The lowest energetic visible photons (red) have more than half of the blue photon energy. So when a material gets excited by the blue photon and releases the red photon energy, what is left is able to generate photon in the IR (so no light) or dissipate the rest as heat (again no light).
So the higher the energy of the emitted photon, the higher percentage of energy ends up in the visible range, so contributing to the light output.
So from 1W of blue primary light you can get a bit more than 0.5W of red, but closely to 0.75W of green. Now with barely 200lm/W for red and 550lm/W for green it means the same 1W of primary blue can generate 100lm of red but 400lm of green light.
So if you mix your phosphor so it generates low CCT, it means you need a big portion of the power in the red part, so less remains for the green, so you get only lower total efficacy.
If you mix the phosphors for higher CCT, so generate less of the reds and more of the green, your efficacy will boost up, easily 2x (2700K vs 5500K) for otherwise the same technology (the fact there are huge differences among various technologies in use, as well as there are differences in their cost). And that means 2x difference in the required power rating.
Now in real life the 5500K is way too harsh to be acceptable (at least in western world), so the industry has settled for about 4000K, still offering more than 50% better efficacy than 2700K for essentially the same technology (so cost).

usually the lower CCT products of seemingly the same product line do tend to offset the inherent efficacy disadvantage by using more expensive more efficient technologies and use the cheaper ones for the higher CCT products, so on the spec sheet the efficacy performance looks similar.

I was thinking about cost, because the only area in my city without 6500k streetlights or HPS is the highway and some private properties. Everywhere else uses 6500k led or HPS.

Although I also prefer 3000K for street lighting, recently on Halloween night the 2700K LED filament lamp I had in my front light on my house died after 10,460 odd hours running.
I actually replaced it with one of these DECOR lamps, and, OMG, are these a dead ringer for HPS!  I only put it in as a curious test?, but it looks so well I’ve decided to leave it in!