OBSTACLE_DISTANCE: add higher angle resolution and angle offset

[magicrub]

This MAVLink v2 packet allows for flexible obstacle detection from multiple sensors of varying resolution while still being backwards compatible.

increment_f (to override increment):
When forced to send exactly 72 samples, having an integer spacing just didn't make sense. Needed to be float so that the correct angular range was understood

angle_offset:
Same as, and adds to, param "PRX_YAW_CORR". However, if you have two sensors you can send multiple messages covering different areas (at varying resolutions). For example, a 360 degree spinning lidar + long-range forward radar/lidar + wide angle sonar(s). With multiple messages, we can aggregate a better picture and down-scale the resolution to whatever the flight controller can handle in that version. And that will improve as we get fancier flight controllers with more memory Meanwhile, the data is no longer limited by the comms link, just on how to store it once received.

count:
Allow the option to send less than 72 distance values.

[OXINARF]

This is a change to common so it needs to be discussed upstream.

I'm completely in favor of angle_offset, I have my doubts about the real-life usefulness of increment_f, but if you are putting it to use then I'm wrong 🙂

[magicrub]

I have my doubts about the real-life usefulness of increment_f

Very important. @rmackay9 and @jmachuca77 really want it too. The issue is if you have 72 samples with integer degree, what happens if those 72 samples cover 90 degrees? Or 20 degrees? This is to prepare the Proximity library for higher resolution target tracking which is being worked on now..

[magicrub]

Here's the upstream PR mavlink#1142

[magicrub]

This has been merged into upstream. I've replaced the commit on this PR branch to be a cherry-pick from upstream. I'll merge this once it passes travis.

merged in upstream