- all CLI parameters related to dynamic idle alone re-named with the `dyn_idle_` prefix
- when linear throttle scaling is active, the user's set idle value is now correct whether dynamic idle is on or off. Previously, the idle value fell when dynamic idle was activated at the same time as linear throttle scaling.
- enabling dynamic idle no longer causes a deadband at full throttle
- the setpoint throttle value sent to Blackbox does not include the dynamic idle offset
- the throttle value sent to the antigravity and dynamic lowpass code includes throttle scaling, but no other modifiers, to avoid false elevation of the apparent throttle position from dynamic idle and unnecessary transient changes in their filter cutoffs
- Dynamic Idle now uses a modified PI controller during active rpm control phase
- the D factor provides early detection of rapid falls in rpm, e.g. in hard chops. It is filtered heavily. Inadequate `dyn_idle_d_gain` may lead to a transient drop in rpm immediately after cutting throttle. Default is 50.
- the P factor provides fast control over rpm during the active control phase. Too much `dyn_idle_p_gain` may cause oscillation in that phase. Note enough and a slow drop in rpm will be inadequately corrected. Default is 50. Needs to be higher with heavier larger props.
- An integral element does most of the work. It prevents enduring offsets from the set rpm. The I gain is high when increasing responding to low rpm, and slow to release. The slow release makes a huge difference and avoids I oscillation. Not enough `dyn_idle_i_gain` and there may be wobble in rpm during the control phase, or the idle value may rise too slowly; too much may cause wobble. Default is 50. Needs to be higher with heavier larger props.
- The DYN_IDLE debug shows idle P, I and D in debugs 0, 1 and 2. minRps stays in debug 3.
- Interactions between throttle and thrust linear, dynamic idle, throttle scaling and throttle boost have been checked and work as they should.
Define block had its #endif in the wrong place causing most of the code to be excluded if `USE_RX_CC2500_SPI_PA_LNA` wasn't defined.
Also fixes compilation errors for CRAZYBEEF4FR legacy target.
Unified targets weren't affected as they define `USE_RX_CC2500_SPI_PA_LNA` so the error wasn't visible.
In the PR adding P to antigravity, the default value was changed so that zero now meant no antigravity, whereas before 1000 meant none. Unfortunately, `ITERM_ACCELERATOR_GAIN_OFF` remained at 1000. This was an oversight.
Note also that the default value of 3500 now results in more antigravity effect than previously; it is equivalent to 4500 in the previous system.
