[token-2022] allow split confidential transfer instruction constructors to be invoked on chain

[samkim-crypto]

Problem

The constructors for the split confidential transfer and confidential transfer with fee instructions do not require proofs to be generated and only require context state accounts. However, these functions are labeled to target only offline usage.

Summary of changes

Remove the restriction on the target build for these functions.

[joncinque]

Looks good! I took a look through the rest of the file, and is there a reason for any of the instruction creators be gated on not(target_os = "solana")? Since none of them are performing any encryption or decryption (that I can see at least), they all seem safe.

Some of them call into the verify_* on the proof program, which will fail, but we could likely relax that restriction in the future. What do you think?

[samkim-crypto]

Looks good! I took a look through the rest of the file, and is there a reason for any of the instruction creators be gated on not(target_os = "solana")? Since none of them are performing any encryption or decryption (that I can see at least), they all seem safe.

Some of them call into the verify_* on the proof program, which will fail, but we could likely relax that restriction in the future. What do you think?

Oh yes, you are right. All of them take in the proof data as input; the proof is not generated inside those functions. It seems like we can remove it. Let me merge this PR and then do it in a follow-up since there might be something that we are overlooking.