I just upgraded to the new WatchCat 2.0.0 and it looks like the "sawtooth sweep" artifacting is back, appearing at the beginning of every output file and repeating approximately every 20 seconds. Previously, on version 1.2.13 and earlier versions, this would only occur if I had more processing threads allocated (under Settings > Performance) than physically available in the hardware. However in this version even with Maximum processing threads limited to 1 (and regardless of the 'one thread per file' or 'multiple threads for one file' optimization setting), it happens with every file processed. This is processing FLAC 16- and 24-bit input to FLAC output with no sample rate conversion, queue input from a Watchfolder.
I have also tried reloading the Stereo Tool licence key in case this was some kind of trial mode limitation.
I have also removed and re-loaded the .sts preset file, which is the same one I was using with 1.2.13.
The StereoTool Preset file is optimized for my batch processing, and does not have any FM/MPX composite generation configuration.
To isolate the issue, I generated 1 minute of silence and encoded into a 16-bit, 44.1kHz .flac file. Then placed this .flac file into a watch folder monitored by WatchCat. WatchCat processed this file and output a new .flac file with the artifacting. So this appears to be independent of source content or dynamics.
I have since rolled back to 1.2.13 for now, and batch processing is working again with the same parameters as before, and also if I increase the processing thread limit back to 8 for fastest performance.
System is a Windows 10 Pro for Workstations x86_64 version 2004 build 10.0.19041.804, processor is a Xeon E3-1535M v6.