Update ckzg4844 to latest version of das branch

[jtraglia]

What type of PR is this?

Other

What does this PR do? Why is it needed?

• ckzg4844.Cell is now an array of bytes, so the conversion is simpler.
• Use fieldparams.NumberOfColumns instead of ckzg4844.CellsPerExtBlob.
• Remove use of ckzg4844.CellsPerExtBlob since it's no longer a public constant.
• We removed the rowIndices params & pass a commitment for each cell.
• We renamed "cell ID" to "cell index"; updated some vars to reflect that.
• Update load trusted setup to support the three types of bytes.
• Pull in trusted setup from the specs which has g1 monomial points.

Other notes for review

I think it's best to review this split-view. The diff looks pretty ugly.

[kevaundray]

This shim file was intentionally made to not rely on other parts of the code -- Can you define NumOfColumns in this file as an alias for CellsPerExtBlob?

The idea is that this file and code by itself can be tested and checked to be correct in isolation to the rest of the code

[jtraglia]

If the length only exists in that package, I don't think it makes sense to have it at all. I've pushed a commit which changes CellsAndProofs to use slices instead of arrays. Open to changing it though.

[kevaundray]

I think it ideally gets checked by the caller who knows fieldparams.NumOfColumns -- There should probably be a check/test that breaks if ckzg.CellsPerExtBlob != fieldparams.NumOfColumns, since these values come from two different sources but should be the same.

If c-kzg for example, were to return less cells than the client code is expecting due to the params, then this would be caught by that test.

RE changing to slice, imo ideally this PR just updates the dependency and other changes which are not necessary are done in a separate PR.

[jtraglia]

I don't like making this change in this PR either. But it was using a constant which no longer exists. ckzg.CellsPerExtBlob isn't available anymore, it's been made private.

[kevaundray]

Oh I didn't realise that CellsPerExtBlob was now private from ckzg -- whether that should be private is out of scope of this PR, so this change makes sense in that context.

[nalepae]

This test fails.

panic: runtime error: slice bounds out of range [:2] with length 0

goroutine 4283 [running]:
github.com/crate-crypto/go-kzg-4844.trim0xPrefix(...)
	/Users/manu/go/pkg/mod/github.com/crate-crypto/go-kzg-4844@v0.7.0/trusted_setup.go:183
github.com/crate-crypto/go-kzg-4844.parseG2PointNoSubgroupCheck({0x0?, 0x0?})
	/Users/manu/go/pkg/mod/github.com/crate-crypto/go-kzg-4844@v0.7.0/trusted_setup.go:110 +0x218
github.com/crate-crypto/go-kzg-4844.parseG2PointsNoSubgroupCheck.func1(0x49)
	/Users/manu/go/pkg/mod/github.com/crate-crypto/go-kzg-4844@v0.7.0/trusted_setup.go:167 +0x64
created by github.com/crate-crypto/go-kzg-4844.parseG2PointsNoSubgroupCheck in goroutine 23
	/Users/manu/go/pkg/mod/github.com/crate-crypto/go-kzg-4844@v0.7.0/trusted_setup.go:166 +0x68
FAIL	github.com/prysmaticlabs/prysm/v5/beacon-chain/blockchain/kzg	0.306s
FAIL

[kevaundray]

Crypto(kzg.go) changes look good to me (Can't explicitly approve)

[nalepae]

All green, thanks a lot!