feat: add support for connection_timeout

[citizen-stig]

So down server can be detected sooner and can be distinguish from slow server

[niklasad1]

@lexnv can you or someone at parity take a look at this PR and #1642 when you have time?

Both looks straightforward to me, #1642 makes the DNS resolution async instead of blocking.

[citizen-stig]

Hey @lexnv , thanks for the quick review! Would be great to have a second review too.

I've fixed formatting. Failed tests look unrelated, this seams to be the root cause

Root cause: serde 1.0.228 split core traits into the new serde_core crate, which changed rustc's diagnostic text. The committed snapshot proc-macros/tests/ui/incorrect/rpc/rpc_empty_bounds.stderr:5 expects the trait for<'de> Deserialize<'de> is not implemented but rustc now emits the trait for<'de> serde_core::de::Deserialize<'de> is not implemented. That one line is the only difference (verified by diffing EXPECTED vs ACTUAL in the CI log).

• Pre-existing on master: the CI run for server: expose subscription id on PendingSubscriptionSink #1634 (merged May 22) fails identically. Unrelated to PR feat: add support for connection_timeout #1643, which only touches client/http-client and CHANGELOG.md.
• Repo precedent for the fix: merged PR Fix CI: update UI test expected result #1423 "Fix CI: update UI test expected result".
• Local Cargo.lock (gitignored) already resolves serde_core 1.0.228 — same as CI — so a local re-bless reproduces CI's output.

Please let me know if anything else needs to be done to get this merged

[niklasad1]

Please let me know if anything else needs to be done to get this merged

It's unrelated to this PR, most likely the UI tests changed slightly because of a rustc toolchain update.

Either just fix the UI tests in this PR or a separate PR, should just be to update tests/ui/incorrect/rpc/rpc_empty_bounds.rs no big deal really.

[citizen-stig]

Hey @DenzelPenzel , thanks for the quick review!
What needs to happen for this PR to be merged?

[idiibrahimggabal41-afk]

Dig nice