the length parameter stays the same when you modulate the start position … start and length determine the playback window of the sampler. When start + length cross the end of the sample then the selection flips around, that is how the logic was designed. It doesn’t “clip” to the end of the sample. So none of that is a bug.
Modulation of start and length resulting in garbled sound is a bug, and I think someone else already reported that one. It must have creeped in in one of the previous updates. I’ll fix this as soon as I can in one of the upcoming updates.
the modulation signals are added on top of the parameter value, so it depends on what value your parameter is set at and what your modulation signal is. For example if the start is at 0.5 and the modulation signal goes from min to max on the scope (full range) then you will have -0.5 … +0.5 on top of that parameter value, so it goes from start to end. However if your start value is higher or lower and your modulation signal is still full scale then yes, it will wrap around, and no, that is not a bug.
you can, you just have to set your start parameter to 0.5 and use a sawtooth modulation signal going from minimum to maximum amplitude.
the sampler does not do time stretching. So if you feed it a modulation signal for the start position and then crank up the signal frequency of course at some point it will start to sound “lo-fi”.
read what I wrote above about the logic of the start/length and how the two work together and react when start+length reaches the end of the sample or when start goes negative from the start of the sample.
There is no point in writing “my sound is f*cked” five times in your posts. Just keep it civilized and ask the questions you have to understand the logic of how the sampler works, then post what is indeed a bug as a seperate bug report, afterwards.