Allow additional/override of non-spec fields, including "address"

[peterbroadhurst]

As discussed in #69:

• The address field is not part of the Version 3 spec - was removed here justification:

  Address unnecessary and compromises privacy.

• The address field is a very Ethereum specific way to store a piece of metadata that is intended to identify the private key publicly
  • An 0x prefixed 20 hex-byte compressed identifier of the SECP256K1 derivied public key, for the private key
• When this library is used to stored key materials for other curves/cryptogrpahy (as now supported since Expose API to directly call HashStruct from EIP-712 and M+N dims #68) its misleading

So...

This PR allows WalletFile.Metadata() to be called get a map[string]any that you can used to:

1. store arbitrary additional fields, like bjjIdentifer or btcIdentifier that are different public key representations of the private key
2. remove the address field from the serialization by setting it to nil

[jimthematrix]

LGTM, approved with a suggestion that can be pushed off to a followup PR.

[jimthematrix]

would be great to have this removed as well, to make the walletFile object agnostic to the public key algorithms. this was done in #69. but no need to hold the merge for this.

[peterbroadhurst]

Yes - agreed... I wonder if I can make it calculate-on-demand as a way to keep the interface identical 🤔

[peterbroadhurst]

great suggestion @jimthematrix - I've managed to do that I think completely removing the field, and even actually removing the Address struct field.

I realized we don't expose the field directly anyway.

Mind taking another look?

[jimthematrix]

thanks @peterbroadhurst , think this is the best we can do given that KeyPair() *secp256k1.KeyPair was part of the package API already. LGTM