PID: change to float input/output

this makes the PID library a bit more flexible for smaller range
numbers. Note that this library is used on ArduPlane and Rover, not
Copter

libraries/PID/PID.cpp

@@ -19,7 +19,7 @@ const AP_Param::GroupInfo PID::var_info[] PROGMEM = {
     AP_GROUPEND
 };
 
-int32_t PID::get_pid(int32_t error, float scaler)
+float PID::get_pid(float error, float scaler)
 {
     uint32_t tnow = hal.scheduler->millis();
     uint32_t dt = tnow - _last_t;

libraries/PID/PID.h

@@ -36,15 +36,11 @@ public:
     /// Positive error produces positive output.
     ///
     /// @param error	The measured error value
-    /// @param dt		The time delta in milliseconds (note
-    ///					that update interval cannot be more
-    ///					than 65.535 seconds due to limited range
-    ///					of the data type).
     /// @param scaler	An arbitrary scale factor
     ///
     /// @returns		The updated control output.
     ///
-    int32_t        get_pid(int32_t error, float scaler = 1.0);
+    float        get_pid(float error, float scaler = 1.0);
 
     /// Reset the PID integrator
     ///
@@ -108,11 +104,11 @@ private:
     AP_Int16        _imax;
 
     float           _integrator;///< integrator value
-    int32_t         _last_error;///< last error for derivative
+    float           _last_error;///< last error for derivative
     float           _last_derivative;///< last derivative for low-pass filter
     uint32_t        _last_t;///< last time get_pid() was called in millis
 
-    int32_t         _get_pid(int32_t error, uint16_t dt, float scaler);
+    float           _get_pid(float error, uint16_t dt, float scaler);
 
     /// Low pass filter cut frequency for derivative calculation.
     ///