Add availableDataOriginalSharesUsed to Header

[Wondertan]

Specs state that availableDataOriginalSharesUsed have to be part of the Header. Current Header implementation misses that

[liamsi]

@adlerjohn would availableDataOriginalSharesUsed according to the spec is

The number of shares used in the original data square that are not tail padding.

Should we extend ComputeShares() to return an int that tells the caller how many shares are actually tailShares s.t. it can be used to set availableDataOriginalSharesUsed in the header?

And how important is it to subtract the len(tailShares) from availableDataOriginalSharesUsed? Isn't it mostly an optimization?

[adlerjohn]

Should we extend ComputeShares() to return an int that tells the caller how many shares are actually tailShares s.t. it can be used to set availableDataOriginalSharesUsed in the header?

Not a bad idea.

how important is it to subtract the len(tailShares) from availableDataOriginalSharesUsed? Isn't it mostly an optimization?

It is an optimization, but it enables potentially significant savings. Since block sizes grow by multiples of 4, it can often be the case that there's a bunch of empty space in the original data square. Up to ~3/8th of the EDS can be "known" and thus doesn't have to be sampled or have row erasure coding computed.