LFPar explodes when freq > nyquist*2

[mtmccrea]

LFPar explodes when freq > nyquist*2.

Environment

• SuperCollider version: 3.13.0-dev
• Operating system: Mac

Steps to reproduce

(
// control rate nyquist
var nyq = s.sampleRate / s.options.blockSize / 2;

{ LFPar.kr(nyq) }.plot(0.2);    // nyquist, ok
// { LFPar.kr(nyq+1) }.plot(0.2);   // a bit over, ok
// { LFPar.kr(nyq*2) }.plot(0.2);   // double, ok
// { LFPar.kr(nyq*2.001) }.plot(2); // beyond double, bye bye...
// { LFPar.kr(nyq*3) }.plot(0.2);   // extreme ...
// { SinOsc.kr(nyq*3) }.plot(2);  // SinOsc, e.g., behaves
)

Expected vs. actual behavior

I would expect LFPar handles aliased frequencies without exploding.

To be tested

[ ] Negative frequencies
[ ] Related UGens that may share the underlying synthesis algorithm with LFPar