Are we talking invariable selection of a number for each faith which varies according to census within a fixed number like inherited peers or invariable selection of a variable number?
But that is a flawed system, as presumably you'd be using religious leaders to represent public census responses. That is inappropriate as you'd be using the most committed religious individuals (the religious leaders) to represent a public who are mainly not actively religious even if they tick a box on the census.
Christianity is the most extreme example - of those that ticked christian on the census about 90% aren't active at all - never attend church, wouldn't consider themselves as a member of a particular denomination and the churches themselves do not consider these people to be their members. In what way does the leader of a particular religious organisation represent people who have chosen not to be active members of that organisation.
So if you really want to follow the census (even allowing for its flawed methodology), for your new proposition of Lords 'life belief' or whatever non-sense term you've come up with, you'll need proportionately (to represent the population).
50% non active 'census' christians
25% non religious
8% active christian - divided between who knows how many denominations
5% muslim (but this needs also to divide between active and non active muslim representation and to represent the various branches of islam)
etc, etc
Complete non-sense and good luck with identifying the organisations and their leaders to represent the well over 75% of the population who are non religious or non active and not members of any religious organisation.
