There were 2 bugs leading to wrong result in Twisted Edwards MSM in #390.
The equality checks should special case neutral elements or a neutral element will be considered equal.
In MSM, we should initialize buckets to the neutral element not via zeroMem which lead to invalid computation for Twisted Edwards curves. This fixes #390. Unfortunately it doesn't seem to help for #366
There were 2 bugs leading to wrong result in Twisted Edwards MSM in #390.
zeroMem
which lead to invalid computation for Twisted Edwards curves. This fixes #390. Unfortunately it doesn't seem to help for #366