The CHLAUTH rules and MCAUSER mappings are only evaluated once when the channel starts, so there is zero impact to a running channel. As part of channel negotiation the additional startup time is negligible. I suppose if you added enough rules you might be able to make a measurable difference but I don't know what that upper limit is.
We currently have a similar situation in that setmqaut rules also match at multiple levels based on precedence. I have customers with thousands of queues and matching setmqaut rules and the additional overhead of opening a queue and evaluating all those rules is negligible. I would expect the timings of evaluating CHLAUTH rules would be similar.
The other point to note is that the performance gains with v7.1 channels will far outweigh the startup cost, even if startup cost were extended by seconds, provided the channels are not constantly being restarted. If you had a client app that was opening a new channel for every message you might notice a difference but I'd recommend rewriting the app rather than avoiding CHLAUTH.