This brings in many of the changes from the PX4 fork on ATLFLight.

Signed-off-by: Mark Charlebois <charlebm@gmail.com>

Also, if the battery is a 3DR Solo battery, it checks for the button
press on the battery which sets the is_powering_off flag.

The check for whether the battery is a 3DR Solo battery is based on
whether the Manufacturer Name, Device Name and Device Chemistry all
match the that of the standard 3DR Solo battery.

Functions added for reading out the following info:
 - Manufacturer Name
 - Device Name
 - Serial Number
 - Device Chemistry

The button press detection relies on reading the MANUFACTURER_DATA
register. The implementation of what information is returned when reading
this register differs from manufacturer to manufacturer.

This has been commented out until an elegant solution can be found to detect
the battery shutdown on different batteries manufactured by different
manufacturers.

This is a cherry-pick of commit 8bd17a4b0cd9c1432cb57a9a80a215692f532370
in 3drobotics/PX4Firmware-solo

Conflicts:
	src/drivers/batt_smbus/batt_smbus.cpp
	src/drivers/drv_irlock.h
	src/lib/ecl

Also fixes the the PEC calculation for writing. See
http://cache.freescale.com/files/32bit/doc/app_note/AN4471.pdf and
http://www.ti.com/lit/an/sloa132/sloa132.pdf for more details for more
details on SMBus reading and writing including with and without PECs.

temporarily use indicated airspeed for transitions so we're consistent with tecs until we move tecs to true airspeed

- created enum for vtol types
- minor cleanup and fixes

- use the transition trust when waiting for TECS to kick in
after transition

- do not run TECS for VTO which are in rotary wing mode
- reinitialise TECS the first time we start using it again

VTOL/TECS: publish transition throttle after transition until tecs is initialized and fw pos ctl is publishing att setpoint

- do not run TECS for VTO which are in rotary wing mode
- reinitialise TECS the first time we start using it again

This change fixes a bug where hash-aware QGC using the old spec would continuously request the parameter list every time it received a mismatched hash.

 - When listing all params, lead with _HASH_CHECK value for cache load
 - When set value on _HASH_CHECK is rx'd, stop any ongoing listing req

the uncostrained demanded pitch was never initialized which
could lead to large initial pitch integral values