Additional (scalar) helper functions.

[astewart-xmos]

I've repeatedly found myself needing to write helper functions such as the following:

int32_t ashr32(int32_t x, right_shift_t shr){
  int32_t y;
  if(shr >= 0) y = (x >> ( shr) );
  else         y = (x << (-shr) );
  return y;
}

int64_t ashr64(int64_t x, right_shift_t shr){
  int64_t y;
  if(shr >= 0) y = (x >> ( shr) );
  else         y = (x << (-shr) );
  return y;
}

int32_t float_s32_to_fixed(float_s32_t v, exponent_t output_exp){
  right_shift_t shr = output_exp - v.exp;
  return ashr32(v.mant, shr); 
}

int32_t float_s64_to_fixed(float_s64_t v, exponent_t output_exp){
  right_shift_t shr = output_exp - v.exp;
  return ashr64(v.mant, shr); 
}

int32_t float_to_fixed(float x, exponent_t output_exp){
  float_s32_t y = f32_to_float_s32(x);
  return float_s32_to_fixed(y, output_exp);
}

Including within lib_xcore_math. I believe there are already macros for the arithmetic shifts, but they aren't exposed as part of the public API. These should probably be added.