Upgrade Recipes to Substrate 3.0

[JoshOrndorff]

This Pull Request updates the Substrate Recipes to use the published Substrate 3.0 crates along with other ecosystem updates such as parity-scale-codec, substrate-wasm-builder, and substrate-fixed.

All code is compiling and passing tests locally. We just have to get CI sorted out. Thanks @jimmychu0807 and @sacha-l for the help.

Checklist:

• Make all pallets compile
• Make all runtimes compile
• Make all nodes compile
• Make all tests pass
• General code quality review and cleanup
• Fix links
• Update writeups

[JoshOrndorff]

@jimmychu0807 Basically this is what I need your help with.

https://github.com/JoshOrndorff/recipes/blob/joshy-3.0/pallets/ocw-demo/Cargo.toml#L13-L14

alt_serde = { version = "1", default-features = false, features = ["derive"] }
serde_json = { version = "1", default-features = false, git = "https://github.com/Xanewok/json", branch = "no-std", features = ["alloc"] }

Ohh! Maybe this will help serde-rs/json#606

[TriplEight]

Looking at the other PRs in this repo I see it's possible to compile Substrate on those Gihub's subtle virtual machines, so I think the reason is in how do you test.
Consider testing the similar way as in substrate

[jimmychu0807]

wow @JoshOrndorff what a feat! Thanks for the Recipe update.

Just two minor suggestion on a few sentences. Also, is substrate-fixed API changed that our code doesn't work with their latest stable release?

I think you can merge the recipe code once you accept the suggestions. I would prefer to look into the CI issue in another time.

[jimmychu0807]

I believe this:

• close Tests recipie not working in v3 #418
• close Missing type for accuracy #415

[TriplEight]

I've fixed CI improving the caching. Previous setup was conflicting with "run in a container" setup.
However, the testing itself takes a whopping 25 min in GHA (and on my testing machine it took under 4 minutes), this might be a sign to use a serious CI system.

[jimmychu0807]

thanks @TriplEight for fixing the CI

[TriplEight]

I wonder why since after 8fc5203 the CI jobs became duplicated.
Before:

after:

Ran an experimental branch from a fork and it was not the case. #431

Maybe it happened due to I committed to this PR directly (and Pr was from a @JoshOrndorff 's fork)

[jimmychu0807]

@TriplEight Let see in future PRs if these CI workflow resume to be normal or run in a duplicated fashion. Don't worry about it too much for now.