Second set of new device support, features and cleanup for the 4.3 cycle.

As we had a lot of tools and docs work in this set, I have broken those
out into their own categories in this description.

New device support
* TI opt3001 light sensor
* TXC PA12 ALS and proximity sensor.
* mcp3301 ADC support (in mcp320x driver)
* ST lsm303agr accelerometer and magnetometer drivers (plus some st-sensors
  common support to allow different WHOAMI register addresses, devices with
  fixed scale and allow interrupt equiped magnetometers).
* ADIS16305, ADIS16367, ADIS16445IMUs (in the adis16400 driver)
* ADIS16266 gyro (in the adis16260 driver)
* ADIS16137 gyro (in the adis16136 driver)

New functionality
* mmc35240 DT bindings.
* Inverse unit conversion macros to aid handing of values written to sysfs
  attributes.

Core cleanup
* Forward declaration of struct iio_trigger to avoid a compile warning.

Driver cleanup / fixes
* mxs-lradc
  - Clarify which parts are supported.
  - Fix spelling erorrs.
  - Missing/extra includes
  - reorder includes
  - add datasheet name listings for all usable channels (to allow them
    to be bound by name from consumer drivers)
* acpi-als - add some function prefixes as per general iio style.
* bmc150_magn - replace a magic value with the existing define.
* vf610 - determine possible sample frequencies taking into account the
  electrical characteristics (defining a minimum sample time)
* dht11
  - whitespace
  - additional docs
  - avoid mulitple assignments in one line
  - Use the new funciton ktime_get_resolution_ns to cleanup a nasty trick
    previously used for timing.
* Fix all drivers that consider 0 a valid IRQ for historical reasons.
* Export I2C module alias info where previously missing (to allow autoprobing)
* Export OF module alias info where previously missing.
* mmc35240 - switch some variables into arrays to improve readability.
* mlx90614 - define some magic numbers for readability.
* bmc150_magn
  - expand area locked by a mutex to cover all the use of the
    data->buffer.
  - use descriptive naming for a mask instead of a magic value.
* berin2-adc
  - pass up an error code rather that a generic error
  - constify the iio_chan_spec
  - some other little tidy ups.
* stk8312
  - fix a dependency on triggered buffers in kconfig
  - add a check for invalid attribute values
  - improve error handling by returning error codes where possible and
    return immediately where relevant
  - rework macro defs to use GENMASK etc
  - change some variable types to reduce unnecessary casting
  - clean up code style
  - drop a local buffer copy for bulk reads and use the one in data->buffer
     instead.
* adis16400 - the adis16448 gyroscope scale was wrong.
* adis16480 - some more wrong scales for various parts.
* adis16300 - has an undocumented product id and serial number registers so
  use them.
* iio_simple_dummy - fix some wrong code indentation.
* bmc150-accel - use the chip ID to detect the chip present rather than
  verifying the expected part was there.  This was in response to a wrong
  ACPI entry on the WinBook TW100.
* mma8452
  - fix _get_hp_filter_index
  - drop a double include
  - pass up an error code rather than rewriting it
  - range check input values to attribute writes
  - register defs tidy up using GENMASK and reordering them to be easier to
    follow.
  - various coding style cleanups
  - put the Kconfig entry in the write place (alphabetically).

Tools related
* Tools cleanup - drop an explicity NULL comparison, some unnecessary braces,
  use the ARRAY_SIZE macro, send error messages to stderr instead of dropping
  them in the middle of normal output.
* Fix tools to allow that scale and offset attributes are optional.
* More tools fixes including allowing true 32bit data (previously an overflow
  prevented more than 31bits)
* Drop a stray header guard that ended up in a c file.
* Make calc_digits static as it isn't exported or in the header.
* Set ci_array pointer to NULL after free as a protection against non safe
  usage of the tools core code.  Also convert a double pointer to a single
  one as the extra level of indirection was unnecessary.

Docs
* DocBook introduction by Daniel Baluta.  Glad we are beginning to
  draw together some more introductory docs to suplement the various
  tools / examples.
* Drop bytes_per_datum sysfs attribute docs as it no longer exists.
* A whole load of missing / fixing of kernel-doc for the core of IIO.
* Document the trigger name sysfs attribute in the ABI docs.
* Minor typos in the ABI docs related to power down modes.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2
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=sCOd
-----END PGP SIGNATURE-----