Fuzz the share commitment correctness

[evan-forbes]

Previous to #1690, we were recalculating the share commitment in each PFB using only subtree roots of the data square

celestia-app/app/process_proposal.go

Lines 170 to 184 in 54b65f9

	for i, shareIndex := range wrappedTx.ShareIndexes {
	commitment, err := inclusion.GetCommitment(cacher, dah, int(shareIndex), shares.SparseSharesNeeded(pfb.BlobSizes[i]))
	if err != nil {
	logInvalidPropBlockError(app.Logger(), req.Header, "commitment not found", err)
	return abci.ResponseProcessProposal{
	Result: abci.ResponseProcessProposal_REJECT,
	}
	}
	if !bytes.Equal(pfb.ShareCommitments[i], commitment) {
	logInvalidPropBlock(app.Logger(), req.Header, "found commitment does not match user's")
	return abci.ResponseProcessProposal{
	Result: abci.ResponseProcessProposal_REJECT,
	}
	}
	}

given the importance of this property and the complexity of the block construction/share encoding, we should replicate this check using some form of fuzzer.

[evan-forbes]

also see #1726 which adds a test for the share commitment

[cmwaters]

I can pick this up. I had thought about appending it to the existing fuzzy test I had written

[evan-forbes]

I had thought about appending it to the existing fuzzy test I had written

that would be perfect!