Fix segment reconstruction

[nazar-pc]

This time bug is really funny and here is what was happening:

• Piece is missing on DSN, reconstruction kicks in
• Reconstructor (unnecessarily, see second commit) prefers source pieces (which works in many cases, but not always, it is async after all)
• If all source pieces are successfully acquired, reconstructor doesn't recover pieces (only commitments and witnesses from source pieces)
• Despite retrieving all valid pieces we end up with a broken reconstructed piece

Reconstructor had an optimization that had incorrect assumptions. The funny thing is that it always worked if source piece was actually missing, but as long as input for reconstruction contained 100% source pieces things didn't work basically at all. Test wasn't checking this unexpected edge case, during DSN sync we never needed to reconstruct parity pieces.

The fix is to simply remove unnecessary condition checking for all source pieces being present (ignore whitespaces to see how small and trivial diff is).

Fixes #1837

Code contributor checklist:

• I have read, understood and followed contributing guide

[nazar-pc]

Test failed due to grandinetech/rust-kzg#236 that had 2 issues combined, one of which incorrect usage of unwrap (again). I'll wait for them to fix it or will get there myself if they don't in a reasonable amount of time.