Fix filterContained

[mitmotw]

What type of PR is this?
Other

What does this PR do? Why is it needed?

filterContained is checking 'contains' between the 'last element of filtered' and new attestation from given attestation list.
So what I'm worrying is like below case.

let the given attestation list has 3 attestations and aggregation bits are like below.
(Just assumed aggregation bits are 6 bits for convenience)

attestation[0].AggregationBits = [1, 1, 1, 1, 0, 0]
attestation[1].AggregationBits = [0, 0, 0, 0, 1, 1]
attestation[2].AggregationBits = [0, 0, 0, 1, 0, 0]

Here I expect all the 3 elements will be included in the final returned list, filtered.

At the first iteration,
attestation[1] will be appended because attestation[0].AggregationBits does not contain attestation[1].AggregationBits.

At the second iteration,
attestation[2] will be appended because attestation[1].AggregationBits (which is the final element of filtered) does not contain attestation[2].AggregationBits.

Like this, all 3 elements will be included in filtered list even though attestation[0].AggregationBits contains attestation[2].AggregationBits.

So this fix stores coveredBits for holding the bits covered so far during the iteration, and
use this for checking 'contains' instead of the 'last element of filtered'.

Which issues(s) does this PR fix?
#12727

[prestonvanloon]

The changes look good to me, but I am curious how other attestation aggregate tests are not impacted by this change

[rauljordan]

Thanks for the contribution @mitmotw !! We're very grateful. As Preston mentioned, here is the gosec failure:

[/home/runner/work/prysm/prysm/proto/prysm/v1alpha1/attestation/aggregation/attestations/maxcover.go:291] - G104 (CWE-703): Errors unhandled. (Confidence: HIGH, Severity: LOW)
    290: 		filtered = append(filtered, al[i])
  > 291: 		coveredBits.NoAllocOr(candidates[i], coveredBits)
    292: 	}



[/home/runner/work/prysm/prysm/proto/prysm/v1alpha1/attestation/aggregation/attestations/maxcover.go:280] - G104 (CWE-703): Errors unhandled. (Confidence: HIGH, Severity: LOW)
    279: 	coveredBits := bitfield.NewBitlist64(candidates[0].Len())
  > 280: 	coveredBits.NoAllocOr(candidates[0], coveredBits)
    281: 

[rkapka]

Hi @mitmotw , I know it's been a very long time since the last activity in this PR. There is a conflict that prevents us from merging it. Can you please resolve it?