There was the case where a param was changed on the Linux side but the
change did not get saved on the DSP side because the param was not in
the local list of changed params on the DSP side. On save, the param
index is now refreshed, and param_get is called on all params that have
changed. This is a hacky workaround but resolves the problem for now.

allows for MAVLink control of safety state

This includes the mtk bugfix for Snapdragon.

NuttX does not inherit file descriptors when creating new tasks.

This test failed on the pixracer because the subscriber thread was too slow
and thus orb messages got lost. This behavior is expected, but the test
should not fail because of that, so we increase the sleeping time.

We now send log messages over mavlink in case excessive vibrations are
detected.

This brings the latest renamed counters.

This needed the set function which sets elapsed to change the name to
avoid ambiguities.

- Make use of the FIFO time offsets provided by the MPU9250 driver.
  It allows to use proper dt for the integration.
- Got rid of the unnecessary perf_counters for now.
- Properly use the changed integrator lib.
- Provide integral_dt for ekf2.

This brings error counters and timestamp offsets between FIFO samples.

The integrator had an untested read mode which did not apply the coning
correction. Instead of keeping two integrals (auto/read) it is now one
and the reset mechanism can be selected by setting the
auto_reset_interval to 0 to disable it or some positive number else.

Also, the integrator could potentially explode if a (single) timestamp
was wrong, so before the last integrated one. This is now caught with a
dt of 0 instead of inf/nan.

This raises the sampling frequency from 500 Hz to 8kHz using the FIFO
buffer.

Before this, I measured a max time spent in poll of 197ms. By checking at
least every 50ms we make sure to not lose any orb messages.

Without that there were still lost messages, now it's all fine (tested
under Linux)