solana: Handle transferring mint authority using SPL Multisig

[nvsriram]

Context:

Currently, there are instructions to safely handle transferring token mint authority to-and-from the NTT Manager - token_authority PDA.

However, with SPL Multisig support, these instructions need to be modified to handle all cases.

List of Abbreviations:

• Multisig(...) = any generic SPL Multisig with signers …
• Multisig(TA, ...) = SPL Multisig with m = 1 and token_authority PDA as one of its required signers
• TA = token_authority PDA
• TA/Multisig(TA, ...) = Representation of NTT Manager
  • Could be token_authority PDA or SPL Multisig with m = 1 and token_authority PDA as one of its required signers

New Proposed Instructions:

Transfer Mint Authority OUT:

• set_token_authority_one_step_unchecked: TA/Multisig(TA, ...) -> any non-signer
• set_token_authority: TA/Multisig(TA, ...) -> non-signer S
  • claim_token_authority: claimed by S as signer
  • claim_token_authority_to_multisig: claimed by S = Multisig(...)
    • ctx.remaining_accounts act as required signers of Multisig(...)
  • revert_token_authority: reverted back to TA/Multisig(TA, ...)

Transfer Mint Authority IN:

• accept_token_authority: signer S -> TA/Multisig(TA, ...)
• accept_token_authority_from_multisig: Multisig(...) -> TA/Multisig(TA, ...)
  • ctx.remaining_accounts act as required signers of Multisig(...)

Implementation Details:

• When the NTT Manager is Multisig(TA, ...): Option<InterfaceAccount<'info, SplMultisig>> is used to avoid additional explicit instructions
• When Multisig(...) needs to sign/validate: ctx.remaining_accounts is treated as required signers

tl;dr:

This PR adds:

• 7 mint authority update instructions to handle all SPL Multisig cases (described above)
• Add NTT namespace TS helpers for the instructions
• Update TS test to cover multiple mint authority transfer instances
• Update IDL

[nonergodic]

Only nits.

One other thing that I noticed that isn't part of the changes here:
The address constraint on the owner signer in RevertTokenAuthority has no custom error, while the other address checks do. I assume this is because usually these type of constraints are checked via has_one which isn't possible here - but I figured I'd flag it here regardless (might be worth a short comment).

[nonergodic]

The code in here could really profit from a bunch of DRY-ness. See e.g. TBRv3 tests that use various helpers under the hood.

[nonergodic]

Suggested change

	additionalSigners: PublicKey[];
	additionalSigners: readonly PublicKey[];

[nonergodic]

Suggested change

	additionalSigners: PublicKey[];
	additionalSigners: readonly PublicKey[];

[nvsriram]

Context:

I refactored out the common functions into claim_from_token_authority and claim_from_multisig_token_authority to avoid repetition. However, I notice admin.rs getting quite lengthy so I would like to refactor it into its own module and then each individual ix (or set of related ix's) can be its own file (or submodule).

Example Refactor:

• set_peer, register_transceiver, set_outbound_limit, and all the other one-off ix's could be their own file.
• transfer_ownership_one_step_unchecked, transfer_ownership, and claim_ownership could be one file.
• accept_token_authority and accept_token_authority_from_multisig could be one file.
• set_token_authority_one_step_unchecked, set_token_authority, claim_token_authority, claim_token_authority_to_multisig, and revert_token_authority could be one file.

Alternative Refactor:

• register_transceiver, set_outbound_limit, and all the other one-off ix's could be their own file.
• transfer_ownership could be its own submodule to separate the "transfer" and "claim" ix's into individual files.
• handle_token_authority could be its own submodule to separate the "accept", "set", and "claim"/"revert" ix's into individual files.

tl;dr:

1. Is there a reason not to refactor admin.rs ix's and instead leave it as one lengthy file?
2. If a refactor is recommended, which option is preferred:
   1. Every ix gets an individual file
   2. Every ix gets an individual file and we use submodules to group related ix's
   3. Every set of related ix's gets an individual file (see example refactor above)
   4. Every set of related ix's gets an individual file but if the set is too large, it gets a submodule (see alternative refactor above)
   5. Variation on one of the above/ None of the above

[nonergodic]

Code is very neat now!
(at least in as far as Anchor code can be neat 😬)

Regarding the refactor:
I think I'd probably have the ownership ixs and the token authority ixs in two separate files, but that's about it.

What I started doing in the Solidity SDK where I'm facing a similar problem of "lots of code, but not actually a lot of complexity" is just put an overview comment at the top of the file that says "look, all this garbage below really just provides the following ...".

As such, it would imho be justifiable to just keep all the code in admin.rs because the file is so entirely uninspiring and unsurprising.

I think you can really just choose here what you like best.