cache the serialization of a Key

This will speed up the common case where a Key is deserialized from
disk, but is then serialized to build eg, the path to the annex object.

Previously attempted in 4536c93bb2
and reverted in 96aba8eff7.
The problems mentioned in the latter commit are addressed now:

Read/Show of KeyData is backwards-compatible with Read/Show of Key from before
this change, so Types.Distribution will keep working.

The Eq instance is fixed.

Also, Key has smart constructors, avoiding needing to remember to update
the cached serialization.

Used git-annex benchmark:
  find is 7% faster
  whereis is 3% faster
  get when all files are already present is 5% faster
Generally, the benchmarks are running 0.1 seconds faster per 2000 files,
on a ram disk in my laptop.

Remote/GitLFS.hs

@@ -344,10 +344,10 @@ sendTransferRequest req endpoint = do
 		LFS.ParseFailed err -> Left err
 
 extractKeySha256 :: Key -> Maybe LFS.SHA256
-extractKeySha256 k = case keyVariety k of
+extractKeySha256 k = case fromKey keyVariety k of
 	SHA2Key (HashSize 256) (HasExt hasext)
 		| hasext -> eitherToMaybe $ E.decodeUtf8' (keyHash k)
-		| otherwise -> eitherToMaybe $ E.decodeUtf8' (keyName k)
+		| otherwise -> eitherToMaybe $ E.decodeUtf8' (fromKey keyName k)
 	_ -> Nothing
 
 -- The size of an encrypted key is the size of the input data, but we need
@@ -355,7 +355,7 @@ extractKeySha256 k = case keyVariety k of
 extractKeySize :: Key -> Maybe Integer
 extractKeySize k
 	| isEncKey k = Nothing
-	| otherwise = keySize k
+	| otherwise = fromKey keySize k
 
 mkUploadRequest :: RemoteStateHandle -> Key -> FilePath -> Annex (LFS.TransferRequest, LFS.SHA256, Integer)
 mkUploadRequest rs k content = case (extractKeySha256 k, extractKeySize k) of