Make Hashcash algorithm configurable

[dahlia]

This patch enables blockchain to consist of blocks with heterogeneous hash algorithms. Few points:

• IBlockPolicy<T>.GetHashAlgorithm(long blockIndex) determines a hash algorithm for a block at blockIndex.
  • Blocks are not aware of which algorithm is used for hashing themselves.
• BlockPolicy<T>() constructor takes an optional hashAlgorithmGetter which configures the above method. If omitted SHA-256 is used for every block in the chain.
  • HashAlgorithmTable.ToHashAlgoirthmGetter() extension method can be useful to describe a simple rule to determine hash algorithms by block indices. E.g., new Dictionary<long, HashAlgorithmType> { [0] = HashAlgorithmType.Of<SHA256>(), [1000] = HashAlgorithmType.Of<SHA512>() }.ToHashAlgorithmGetter() is equivalent to (idx) => idx >= 1000 ? HashAlgorithmType.Of<SHA512>() : HashAlgorithmType.Of<SHA256>().
• HashAlgorithmType is a type for comparing hash algorithms.
  • It can be instantiated using Of<T>() method, like HashAlgorithmType.Of<SHA256>().
  • It implements equality methods and operators. E.g., HashAlgorithmType.Of<SHA256>() != HashAlgorithmType.Of<SHA1>(), HashAlgorithmType.Of<SHA256>().Equals(HashAlgorithmType.Of<SHA256>()). Also can be elements of ISet<T>/IImmutableSet<T> or keys of IDictionary<TKey, TValue>/IImmutableDictionary<TKey, TValue>.
• As proof-of-work does not directly determine Block<T>.Hash, but Block<T>.PreEvaluationHash, BlockHash became fixed-size (32 bytes).
• As various hash algorithms can be used for proof-of-work, Block<T>.PreEvaluationHash became represented as variable-length ImmutableArray<byte> (instead of BlockHash, which was misleading).

[codecov]

Codecov Report

Merging #1352 (a504f62) into main (0033ed7) will decrease coverage by 8.83%.
The diff coverage is 49.55%.

❗ Current head a504f62 differs from pull request most recent head 8dc9f89. Consider uploading reports for the commit 8dc9f89 to get more accurate results

@@            Coverage Diff             @@
##             main    #1352      +/-   ##
==========================================
- Coverage   86.03%   77.19%   -8.84%     
==========================================
  Files         361      256     -105     
  Lines       32457    17257   -15200     
==========================================
- Hits        27925    13322   -14603     
- Misses       2747     3371     +624     
+ Partials     1785      564    -1221     

Impacted Files	Coverage Δ
...bplanet.Explorer/Controllers/ExplorerController.cs	0.00% <0.00%> (ø)
...rer/Controllers/GenericControllerNameConvention.cs	0.00% <0.00%> (ø)
Libplanet.Explorer/Controllers/GraphQLBody.cs	0.00% <0.00%> (ø)
Libplanet.Explorer/ExplorerStartup.cs	0.00% <0.00%> (ø)
Libplanet.Explorer/GraphTypes/NodeStateType.cs	0.00% <0.00%> (ø)
...ibplanet.Explorer/Interfaces/IBlockChainContext.cs	0.00% <0.00%> (ø)
Libplanet.Explorer/Queries/BlockQuery.cs	0.00% <0.00%> (ø)
Libplanet.Explorer/Queries/ExplorerQuery.cs	0.00% <0.00%> (ø)
Libplanet.Explorer/Queries/TransactionQuery.cs	0.00% <0.00%> (ø)
Libplanet.Explorer/Store/AddressRefDoc.cs	0.00% <0.00%> (ø)
... and 268 more

[longfin]

We may need ValidationContext to consolidate the hash algorithm and DateTime... 🤔

[dahlia]

As HashAlgorithmGetter is usually from IBlockPolicy.GetHashAlgorithm, even if we make a new data type its construction would be unnatural IMHO.