MS5611 rework

[nicolaerosia]

Highlights:

• No longer use Work Queue, use a dedicated thread. I need someone using QuRT to review if pthread support is available (@mcharleb)
• The previous logic was flawed, since we were using a fixed callback timer, we were sampling the temperature, waiting half a period (much more than the required delay), then sample the pressure. ** the idea is to align this driver with the one on NuttX so we can merge it later on (they do multiple pressure samples with each temperature sample - we can do that too but I did not want the changes to be so intrusive)
• Fix the invalid sample I was getting, https://github.com/PX4/DriverFramework/pull/198

timespec_inc can be factored out into a header so multiple drivers can use it.

@mcharleb @julianoes @LorenzMeier please review

[dagar]

Likely doesn't matter, but you don't really need to read the temperature constantly. In the ms5525 airspeed driver (nearly identical to this) I only update the temperature every 10th cycle.

[nicolaerosia]

@dagar that is correct, see my PR description "The previous logic was flawed, since we were using a fixed callback timer, we were sampling the temperature, waiting half a period (much more than the required delay), then sample the pressure. ** the idea is to align this driver with the one on NuttX so we can merge it later on (they do multiple pressure samples with each temperature sample - we can do that too but I did not want the changes to be so intrusive) "

You might also find this interesting: http://www.freesteel.co.uk/wpblog/2014/11/26/ms5611-altitude-sensor-very-sensitive-to-own-working-temperature/