c_nonce & credential batches behaviour not clearly stated

When the wallet is requesting (say) a batch of 10 credentials, it could either:

1. Request c_nonce once and use it in all 10 proofs send in the array to the credential endpoint
2. Request c_nonce 10 times and send 10 proofs each with a potentially different nonce in the array to the credential endpoint

I can't find anything that makes a specific direction clear, though it does say in the proof replay section

A Wallet can continue using a given nonce until it is rejected by the Credential Issuer.

I think '1' is the sensible way to do it as I can't see what benefit you can get from having 10 different nonces - a key proof is either fresh or it's not, and 10 identical nonces matching the most recent c_nonce the server has returned is sufficient to guarantee freshness.

If we agree '1' is intended, then I think it's probably sensible to add two clauses somewhere:

When supplying multiple proofs, all proofs MUST have the same c_nonce.

and:

The c_nonce included in the proof MUST be the c_nonce most recently received from the issuer.

We saw behaviour '2' is a wallet that failed the conformance tests. Both the issuer and wallet tests currently expect behaviour '1'.

Start with the proof replay section and the credential endpoint's multiple-proof requirements, then compare them with the conformance tests mentioned in the issue. Trace whether the specification distinguishes one c_nonce reused across a batch from separate nonces. Done means the intended batch behavior and nonce freshness rule are stated unambiguously in the appropriate specification section.