Add an NU5 block test vector that contains many shielded Orchard transactions

[dconnolly]

Motivation

When blocks started showing up on mainnet after NU5 with many shielded Orchard transactions, we observed aberrant zebrad verification behavior that helped us identify some bugs already. We should include these blocks (at least one) as test vectors.

Related Work

#4775

[str4d]

I wrote a script for generating (most of) the necessary data to benchmark verification of specific heavy blocks (the heaviest I could find at their respective times) using a mainnet zcashd, which we use to avoid checking in a large amount of data that is only used for performance benchmarks. The script collects the noted blocks and their transparent inputs (in zcashd's coins LevelDB format), and writes out a tarball. We then pair that with some hard-coded Sapling and Orchard trees (that I extracted by hand, as the RPCs didn't quite have enough expressiveness to expose them) that are needed to fully-validate the blocks.

• https://github.com/zcash/zcash/blob/master/qa/zcash/create_benchmark_archive.py
• https://github.com/zcash/zcash/blob/882935e35ee338876c071f9cd93aa3694154de73/src/zcbenchmarks.cpp#L436-L702

(For reference, the usage of these in our benchmarks is ./qa/zcash/performance-measurements.sh time connectblockslow or connectblocksapling / connectblockorchard)

[teor2345]

Zebra's performance is good enough for now, so this is a low priority.