Added PWM module changes since #4638 was merged

Hmm, this seems still overly complicated to me, why not just make this as part of the GPIO driver, e.g. with some CPU internal function as gpio_init_af/mux or similar everything should be in order?! This is what we did for many other CPUs and I can't imagine that the kinetis are so much different...

I don't see what is overly complicated? The only difference between this and other CPUs is that the port_init function is inside port.c instead of gpio.c and exported so that it can be used from the other periph drivers as well instead of only from inside gpio.c.

ah, and the push/pull configuration. I built this on the premise that we'd need that parameter but depending on the outcome of #4750 and #4472 we might need to refactor port_init a little.

But why put in i'ts own module. I think there is actually no need to export functions like '_gpio_port(gpio_t dev)' or '_gpio_port(gpio_t dev)` to the other periph drivers, as these are only needed internally in the GPIO driver. This keeps the code simpler and IMHO easier to follow.

The idea is that the periph drivers will also use port_init to select their desired alternate functions in the mux. I don't know how to do this without exporting some new symbol or changing the current GPIO API to support an AF parameter?

Here is how I would envision the gpio module to look like: #4830. Only thing open is the open-drain configuration, but can be put in easily... I think #4830 looks much easier (at least to me :-))

so when using this from other peripheral drivers (e.g. spi/uart), you just call

gpio_ini(mosi_pin, GPIO_DIR_OUT, GPIO_NOPULL);  // adapt to changes when integrating OD mode...
gpio_init_af(mosi_pin, GPIO_AFx);

-> easy to use, and no need for the SPI driver to know anything about ports/gpio/pins in depth...

Closing in favour of #4830