Optimized validate_blob_transaction_wrapper() function

[asn-d6]

// Verify that the list of `blobs` map to the list of `commitments`
//
// This is an optimization over the naive approach (written in the EIP) of iteratively checking each blob against each
// commitment.  The naive approach requires n*l scalar multiplications where `n` is the number of blobs and `l` is
// FIELD_ELEMENTS_PER_BLOB to compute the commitments for all blobs.
//
// A more efficient approach is to build a linear combination of all blobs and commitments and check all of them in a
// single multi-scalar multiplication.
//
// The MSM would look like this (for three blobs with two field elements each):
//     r_0(b0_0*L_0 + b0_1*L_1) + r_1(b1_0*L_0 + b1_1*L_1) + r_2(b2_0*L_0 + b2_1*L_1)
// which we would need to check against the linear combination of commitments: r_0*C_0 + r_1*C_1 + r_2*C_2
// In the above, `r` are the random scalars of the linear combination, `b0` is the zero blob, `L` are the elements
// of the KZG_SETUP_LAGRANGE and `C` are the commitments provided.
//
// By re-grouping the above equation around the `L` points we can reduce the amount of scalar multiplications further
// (down to just `n` scalar multiplications) by making the MSM look like this:
//     (r_0*b0_0 + r_1*b1_0 + r_2*b2_0) * L_0 + (r_0*b0_1 + r_1*b1_1 + r_2*b2_1) * L_1
func VerifyBlobs(commitments []*bls.G1Point, blobs [][]bls.Fr) error

Same logic should be used for verify_blobs() on the CL.

[protolambda]

Awesome, Merged into #1 (github doesn't seem to detect the rebase/merge though):

• Rebased onto datablobs
• Cleaned up the Blobs() and Points() methods (Blob is an array type now, and made Blob.Parse() method, and renamed the other point/fr parse methods to Parse()), and reduce allocations
• Updated test to reflect the slice->array type change of Blob, had to fix a mutation on old copy