Consensus calculation algorithm description structure

The way computing-consensus.md is structured is less than ideal. I think it could do with being substantially restructured, to reduce the amount of back and forth reading needed (and to either avoid, or make clearer, situations where one part of the algorithm is overridden by another).

Also, some information about consensus calculations is elsehwere, eg in consensus-formats.md.

And, the handling of conditionality on the consensus method (most of which are presumably hitorical) is very intrusive. IMO the text should principally describe the latest consensus method, and the backward compatibility information should be collected out-of-line (and be relegated to the Historical chapter when it's no longer needed).