modules: hal_nordic: nrfx: use dt_nodelabel_exists

As of today, nrfx components are selectable depending if devicetree has
a certain compatible or not, and in case of IP that can operate in
multiple modes, nodelabel+compatible. An example of the later is:

```
config NRFX_SPI0
    bool "SPI0 driver instance"
    depends on $(dt_nodelabel_has_compat,spi0,$(DT_COMPAT_NORDIC_NRF_SPI))
    select NRFX_SPI

config NRFX_SPIM0
    bool "SPIM0 driver instance"
    depends on $(dt_nodelabel_has_compat,spi0,$(DT_COMPAT_NORDIC_NRF_SPIM))
    select NRFX_SPIM

config NRFX_SPIS0
    bool "SPIS0 driver instance"
    depends on $(dt_nodelabel_has_compat,spi0,$(DT_COMPAT_NORDIC_NRF_SPIS))
    select NRFX_SPIS
```

These symbols are later selected by e.g. SPI driver if `spi0` node is
enabled. While this works in the Zephyr context, it can be a problem in
applications using nrfx directly in the application layer, unless they
tweak devicetree. For example, if devicetree `spi0` node defaults to
`compatible = "nordic,nrf-spim"`, NRFX_SPI0 or NRFX_SPIS0 won't be
selectable, but an application using nrfx directly doesn't really care
about what is defined in devicetree as it is effectively bypassing it.
It just wants to select e.g. `CONFIG_NRFX_SPI0=y` and forget about the
rest.

This patch fixes this problem by just checking for the existence of
certain nodelabels.

Signed-off-by: Gerard Marull-Paretas <gerard.marull@nordicsemi.no>

Author: gmarull