Block root caching in blob sidecar

[realbigsean]

With #4900 we no longer have the block root in a blob sidecar. We sometimes need the block root when we only have a sidecar, especially in RPC when calculating BlobIdentifier.

We should try to figure out how important caching this actually is, and if it's worth the effort of making sure we don't re-calculate the block root for a blob sidecar. Or since we're calculating it per-sidecar, (and again in the actual block), maybe we could cache it across sidecars and check equivalence.

But the SignedBlockHeader is very small, so I'm not sure it'll have a big impact

[realbigsean]

@pawanjay176 checked this out and it looks like calculating the block root from a header takes microseconds so this does not seem worth optimizing