spl: Bump to newest SPL crates for token-2022 release

[joncinque]

Problem

The new SPL crates, including token-2022, have been released, and the new token-2022 program will be deployed soon, but the monorepo is still using the old crates.

Summary of Changes

In preparation of the release, bump the monorepo to the newest crates.

The breaking changes concern extensions that have not been released yet (token-group and confidential transfer instructions). The main impact is that RPC servers on 1.18 will incorrectly parse the token-group type once the new program is deployed. I don't think we should backport this change to 1.18, so we'll need to live with some incorrect parsing until 2.0 hits mainnet.

On the flipside, since token-2022 built with these crates will be deployed to all networks, we should backport this change to 2.0. The next step will be to add the built program to program-test / solana-test-validator and backport that too.

Fixes #

[CriesofCarrots]

Looks like it's just spl-pod that's duplicated now, which seems like because of spl-type-length-value > spl-transfer-hook-interface.
But I won't quibble; it's a lot less bloat now.

[joncinque]

I looked a bit more into why there were two versions of spl-pod, which made me realize that a lot of the packages needed to be republished since they missed another bump. I've done those bumps correctly, and we're back down to one version of everything

[joncinque]

Sorry, I had to rebase to pick up the changes from #2252

[joncinque]

Oops, never mind, this won't be possible until we land #2817 and update SPL to use 2.0 on the newer curve25519 crate. Even with the Cargo.lock issue resolved, the build then fails because the confidential transfer proof generation is using the old curve25519 crate, but the zk-sdk is using the new one.

Either way, I'll move this back to draft while waiting for everything else to get done.

Edit: never mind again, seeing that 2.0 will hit mainnet in a month, we should probably not backport a new major version of curve25519, and instead wait until 2.1

[joncinque]

Closing in favor of #3431