Remove spisvc from the mangOH SDEF since the default.sdef from Legato
18.07.0 includes spisvc and mangOH.sdef includes default.sdef indirectly
via wifi.sdef.

The bq27xxx module now compiles for both 9x15 and 9x07 based targets.
The module has not been tested yet.

Specifically, MANGOH_WP_CHIPSET_9X15 and MANGOH_WP_CHIPSET_9X07
variables have been defined.  This allows for decision making in .cdef
and .adef files based upon which module is being targeted.  This is
useful when requiring libraries and the versions of the libraries differ
between chipsets.

Rather than requiring a yocto re-build to support the CAN IoT card, the
relevant kernel module code has been taken out of tree.  For 9x15 based
modules (wp75, wp85) use the modules in the can_9x15 directory.  For
9x07 based modules (wp76, wp77), use the modules in the can_9x07
directory.  The can_common/can_iot module instantiates an mcp251x device
with the appropriate platform data for using the CAN IoT card in a
mangOH Red.

* Rename .mdef files to remove numeric prefix that was used to force
  kernel modules to load in the required order
* Use new "requires module" feature of .mdef files to specify
  dependencies.
* Update mtwifi script based on module renaming

BMI088 has typo for accel calculation which is incorrect.  Updated
calculation to be correct.

Updated sysfs config show and restore functions to use library calls

Previous commit 354e27dd introduced a dependency upon IIO_GASRESISTANCE
being defined as a member of enum iio_chan_type.  The enum is part of
the kernel headers and not part of the out-of-tree IIO code in the
mangOH repository.  The code now checks that
CONFIG_IIO_CHAN_TYPE_GASRESISTANCE is defined before trying to reference
IIO_GASRESISTANCE.

By using the componentSearch feature, more of the build requirements can
be specified in the SDEF instead of the Makefile.  componentSearch is a
Legato 18.04.0 feature.

The mangOH build is failing in the developer VM based on Ubuntu 18.04
because the wrong toolchain is being selected for the red_wp85 build.  I
think the root cause may have to do with the .phony target that sets up
the Make variables not being run multiple times.  Rather than spending a
lot of time on this, I have just simplified the Makefile so that the
problem goes away.