ekvset is not alignment safe #1
Description
Currently, while \ekvparse is alignment safe, \ekvset isn't.
A fix would be possible by putting \ekv@alignsafe at the start of \ekvset and using \ekv@endalignsafe before and another \ekv@alignsafe after every executed user code. This has an estimated performance hit of more than 10%. (This would also require changes in \ekvchangeset, and maybe a few other places)
Unlike with some other key=val parsers (that have some sort of brace bug), category 4 tokens (&) can be parsed without a problem by putting braces around them, so with a slight user adjustment this can be fixed with ease without affecting every other usage of \ekvset. Also, it is possible to use \ekvset in an alignment safe way (see expkv-cs, which does exactly that). And if the code is fully expandable, and \ekvset is the first thing to appear in an alignment cell, & aren't a problem as well (because TeX then scans for \omit during which it doesn't replace &).
Currently I tend to not make \ekvset alignment safe, but am open for discussion.