Isn’t proof aggregation a solution here? I see in your table that the RISC Zero proof is 2 MB, which would fit inside a current bitcoin block, albeit in a nonstandard way. Although I would propose that an implementation of this should create a new datacarrier (or “proofcarrier”) “sidecar” to each block for aggregated proofs, similar to what @EthanHeilman has proposed with BitZip, published here and presented (by me) here.