Finally, I updated all drivers and their saul adaption when available. In the mean time, I also removed some *_params.h files for some boards because I think they are useless since some values are already defined in the board.h and then are taken in to account by *_params.h in the drivers directory.

Nice effort! I think most of the changes are very valid and its nice to see the style of parameter definition to be cleaned up.

One think I don't like, that this PR is still using (slightly) different styles in the same context. For example looking at the fox board: the parameters for the radio are defined using some kind of static struct initializer style

#define AT86RF2XX_PARAMS    { .spi = SPI_DEV(0),                \
                              ...
                              .reset_pin = GPIO_PIN(PORT_C, 1)}

while the on-board sensors are configured using singular defines

 #define L3G4200D_I2C        I2C_0
 #define L3G4200D_ADDR       0x68
 #define L3G4200D_DRDY       GPIO_PIN(PORT_B,8)
 #define L3G4200D_INT        GPIO_PIN(PORT_B,11)

So I would prefer to keep to one style. IMHO the singular defines are the nicest, as this allows us to use default values which we don't need to re-define. So taking the l3g4200d as example, The fox configuration for this sensor would look like:

/**
 * @name    Define the interface for the L3G4200D gyroscope
 * @{
 */
#define L3G4200D_PARAM_DRDY       GPIO_PIN(PORT_B,8)
#define L3G4200D_PARAM_INT        GPIO_PIN(PORT_B,11)
/** @} */

The I2C dev and addr values then simply fallback to their default values I2C_DEV(0) and 0x68. NOTE: the l3g4200_params.h file needs also to be adapted as it does not define default values currently...

The same would be true for the at86rf radio, whose configuration would look like this:

/**
 * @name    Define the interface to the AT86RF231 radio
 */
#define AT86RF2XX_PARAM_CS          (GPIO_PIN(PORT_A, 1))
#define AT86RF2XX_PARAM_INT         (GPIO_PIN(PORT_C, 2))
#define AT86RF2XX_PARAM_SLEEP       (GPIO_PIN(PORT_A, 0))
#define AT86RF2XX_PARAM_RESET       (GPIO_PIN(PORT_C, 1))

Here again: SPI dev and clock speed use the default values...

Thanks for having a look @haukepetersen.

the singular defines are the nicest, as this allows us to use default values which we don't need to re-define

I totally agree with this and will apply the required change.

After looking at the board params (iotlab-m3, fox, etc), the actual status of this PR is leaving them broken. I'm working on fixing them...

@haukepetersen, I'm nearly done with this one (at least the board.h are ready). Can you have a look again ?

Ok, after some fight, CI is finally green here... but the diff has increased significantly.

@gebart @haukepetersen, can you check that all mulle auto configured devices are still working ?
Is there someone with pba-d-01-kw2x and fox boards that can try ?

It's ok on iotlab-m3.

Maybe @smlng is also interested by this one ?

Closing in favor of all mentioned PRs above and the remaining ones that will come soon

I think I'm done but I'm not 100% sure. I admit I lost myself in all those branches and PRs...

Each one should be simpler to review @kYc0o ;)

But now I'm wondering what was the simplest. It will be a lot of work reviewing all that, right ?