- the SuperBlock already does this

The debug messages are too verbose to be run in a production vehicle and inherently were something that should only be run in SITL / debug sessions on hardware. Switching the flag to the PX4_DEBUG() macro does not only make this more explicit, but also saves a lot of flash space that otherwise was consumed by the strings.

This ensures that users do not accidentally enable this in an airborne system and saves flash space.

And also format file according to code style.

 - fixes #8585

This bound can be quite tight, because:
- The system still behaves well, even if all FD's are used (as opposed to
  a stack overflow)
- The amount of used FD's is typically only increased one at a time
  (e.g. adding new logged topics, adding a mavlink stream, ...)
- reducing CONFIG_NFILE_DESCRIPTORS to the minimum frees up a considerable
  amount of RAM on systems that need it

Makes sure that if CONFIG_NFILE_DESCRIPTORS == 0, no warning is printed.

It is important that tasks do not run out of them and this addition will provide a warning and log evidence if it goes wrong.

See https://github.com/mintty/mintty/wiki/Keycodes#editing-keys

with this commit, lines can be edited with the lef/right arrow key

Simulator: virtual battery drain stops by default at 50% to avoid failing preflight checks to take off again
also added comments and removed throttle compensation because there is no real battery emulation with load drop

Battery: enhanced voltage, capacity estimate fusion, set empty voltage to a useful value for reasonable lipo usage

Battery: adapt output constraints to possible values, only calculate remaining capacity if capacity is configured

Battery: refactor _param out of the variable names and use 0.f for zero floats to make code more readable

Battery: simplify linear voltage relation, currently neglecting the real world voltage drop under load

it's a lot more readable and battery draining stops exactly at the empty battery voltage which makes debuging easier because it's directly predicatable

In rare circumstances, the buffer size was too small by 1 byte, leading
to a param export failure. And leading to a reset of all params!
This could only happen when the last saved param was a float.

Also the realloc_ok initialization is not needed.

It's used with real OS FD's.

Some modules call this (and maybe in rare circumstances), which could lead
to stack overflows. With our current headroom of >= 300bytes, 256 should
be ok.

This increases the export time from ~28ms to ~34ms in my case (on a
Pixracer)