The reason is more or less this: in 99% of cases the two asynchronous parameters are in sync anyway. So by recording them asynchronously I effectively get twice the sample rate for "elevator position". Only in the rare cases where they don't agree (disconnect pulled or mechanical failure, say) does that not apply. In the majority of THOSE cases, it's clear what is going on - for a jam, for example, you'd see one elevator stuck and one still moving. Add in CVR type evidence (which would confirm the disconnect had been pulled, for example) and you can pretty easily resolve any differences.
On older units, asynchronous recording was a way to capture both the required rates and record all the relevant items, since memory capacity was an issue. It's less so today, but old habits die hard.