SASL PLAIN and anonymous authentication

[Jokler]

This PR fixes the following divergences between the prototype branch and #31:

• a fallback to the "anonymous" route if the SASL capability was not enabled
• a new OrbitError type which is used to return
  The other requirements of SASL PLAIN and ANONYMOUS authentication #31 have already been implemented on the prototype branch but this PR should be used to review any code relevant to SASL PLAIN and ANONYMOUS authentication #31.

Additions after review:

• Improve SASL flow to match the spec better
• Fix bug in chunking
• Bubble up the SASL error text
• Set the network name in ServerMetadata from ISUPPORT
• Log anyhow error chain and bubble up the top level error as a String
• Return sign in User/Guest status after sign_up is called

Resolves: #31

[Jokler]

I'll leave fixing that build error to @dolanske because reproducing it locally is blocked by other errors on my system.

[zealsprince]

Overall I like where this landed. The typed OrbitError is a real improvement over stringly errors at the boundary, and the .context() pass means failures will actually be traceable now. Left inline comments, the two I'd want sorted before merge are the split_at panic and the question of waiting for the AUTHENTICATE challenge. The silent fallback I'd at least want a shared understanding on. The UI surfacing can be a follow-up issue.

A few things that didn't fit inline:

• The lockfile looks half-updated: catalog says vite-plus ^0.2.4 but the catalogs section resolves 0.2.1, root gets 0.2.4, and the apps stay on 0.2.1. Does a fresh pnpm install settle it for you?
• The @ts-ignores on defineConfig, is that a vite-plus 0.2.4 type regression? If so I'd swap to @ts-expect-error with the upstream issue linked so they self-report when fixed.
• The wasm-pack block is copy-pasted between ci.yml and cd.yml and runs cold each time. Composite action plus caching once the sourcing questions are answered?

Curious about the nightly requirement too, is that just for the if-let match guards or is more coming that needs it?

Also, we should look to resolve the CI errors with @dolanske so that we can get a nice clean PR out there :)

[Jokler]

The lockfile looks half-updated: catalog says vite-plus ^0.2.4 but the catalogs section resolves 0.2.1, root gets 0.2.4, and the apps stay on 0.2.1. Does a fresh pnpm install settle it for you?

That printed this, but didn't actually change anything:

devDependencies:
- vite-plus 0.2.1
+ vite-plus 0.2.4

The @ts-ignores on defineConfig, is that a vite-plus 0.2.4 type regression? If so I'd swap to @ts-expect-error with the upstream issue linked so they self-report when fixed.

I don't know what's causing it, I actually updated vite-plus in the hopes of fixing that issue. @dolanske suggested to ignore the problem for now so maybe he can try to look into it once he is available.

The wasm-pack block is copy-pasted between ci.yml and cd.yml and runs cold each time. Composite action plus caching once the sourcing questions are answered?

Sounds good but I don't want to implement that in this PR.

Curious about the nightly requirement too, is that just for the if-let match guards or is more coming that needs it?

There is no nighly requirement, I have removed it. :)

[zealsprince]

lol

[zealsprince]

Good with your feedback! I'll capture the caching concern for the packages as a separate ticket!