Weight-P is optional. It's on by default with "smart" (weightp=2) mode, yes. But you can reduce it to "blind" (weightp=1) mode or even disable it (weightp=0).
Additionally some of the faster presets reduce/disable Weight-P. The "baseline" profile will disable Weight-P too!
CoreAVC only has problems with "smart" mode. And even that only in Non-CUDA mode. Furthermore we shouldn't restrict our encodes, just because one decoder is broken.
Instead we should spread weightp=2 encodes and encourage the decoder developers to get their code fixed asap