feat: introduce observed, observed_macros and observed_testing crates

[Evgenii (Vaiz)]

Move the observed telemetry crate family from the internal ox-sdk monorepo into this repository, so the crates live alongside the rest of the open-source Oxidizer surface and can be published to crates.io. This is the publication step for the telemetry-enrichment work; until now every change had to be transferred between the two repositories by hand.

The crates

• observed (0.24.0) — structured, typed telemetry events. Each event is a struct with a data classification on every field. Enrichment entries are scoped, stackable and propagate across threads and async tasks. One event struct can produce a log and a metric with independent field subsets per signal.
• observed_macros (0.24.0) — the #[event(...)] attribute and #[derive(Enrichment)]. Re-exported through observed; do not depend on it directly.
• observed_testing (publish = false) — the MockProcessor harness and the integration-test suite for both crates.

Changes made during the move

• observed no longer depends on opentelemetry. Value is an owned enum, so an opentelemetry major bump is no longer a breaking change for a crate that only emits typed events. opentelemetry stays as a dev-dependency, because the examples and benchmarks still export through an OpenTelemetry sink.
• The metric_number_of helper moved into observed_testing, which is its only consumer in this repository.
• observed_testing drives futures by hand with testing_aids::FutureTestExt and testing_aids::poll_once instead of the ox-sdk-only helper crate.
• Tests and examples use expect with a message instead of unwrap, to satisfy this workspace's clippy::unwrap_used policy.
• The #![doc(html_logo_url)] and #![cfg_attr(coverage_nightly, ...)] attributes are enabled, matching the convention of the other crates here.
• observed_testing is registered in automation::INTERNAL_CRATES.
• all_the_time and thread_local are added to [workspace.dependencies].

Effects

• Telemetry changes no longer need a manual transfer between two repositories; they also no longer reach internal consumers until a release is published from here.
• Leaf crates that emit typed events but never export to OpenTelemetry stop compiling opentelemetry and its transitive tree; a crate that does export must now convert values itself.
• The public API is governed by this repository's gates — unwrap_used, external-type approval, coverage and mutation testing — which are stricter than the ones the crates were developed under.
• The Value variant set becomes public vocabulary rather than an implementation detail. #[non_exhaustive] keeps adding a variant source-compatible for matchers, but an exporter outside this repository needs a catch-all arm.

Validation

just check, just anvil-clippy, just anvil-fmt, just test (6628 passed), just test-docs, just readme-check, just anvil-license-headers, just anvil-external-types, just anvil-doc-build and cargo machete all pass locally. cargo spellcheck cannot be installed on this machine and mutation testing has not been run locally; both are left to CI.

[github-actions]

⚠️ Manual proc-macro SemVer review required

cargo semver-checks intentionally does not analyse proc-macro-only targets. 1 of 2 publishing crate(s) require manual review either for their own procedural macro contract or because they directly consume a manually reviewed breaking release. Ordinary libraries retain their automated result, and review advances to the next dependency edge only when the current release is breaking.

Crate	Baseline	Baseline commit	This PR	Minimum required	Status
observed	new crate	—	0.24.0	0.24.0	✅ ok
observed_macros	new crate	—	0.24.0	—	⚠️ manual proc-macro review required

⚠️ observed_macros — manual proc-macro review detail

`observed_macros` is a proc-macro-only crate. cargo-semver-checks intentionally skips proc-macro targets because they have no supported library API surface. Review exported macro names, accepted input syntax, diagnostics, and generated output manually; build and test results do not establish public API SemVer compatibility.

Proc-macro API compatibility must be reviewed manually; successful builds and tests do not establish SemVer compatibility. This check is informational and does not block the merge.

View the check run

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 100.0%. Comparing base (9877935) to head (2ba5fcd).

Additional details and impacted files

@@            Coverage Diff            @@
##             main     #671     +/-   ##
=========================================
  Coverage   100.0%   100.0%             
=========================================
  Files         503      532     +29     
  Lines       57406    59660   +2254     
=========================================
+ Hits        57406    59660   +2254     

Flag	Coverage Δ
linux	71.4% <100.0%> (-28.6%)	⬇️
linux-arm	70.7% <100.0%> (-29.3%)	⬇️
scheduled	?
windows	71.4% <100.0%> (-28.6%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[martintmk]

Approved, feel free to ignore my agent.

[martintmk]

I played with the branch a bit and I think this is really nice.

Two nits that we can address in follow-up:

• Support for histogram boundaries. This is something that is encoded in defaults for each histogram and we should preserve this. See MetricsHandler for example.
• Have you thought about using CompactStr in Text? Could help with some small allocations.
• I could see the following pattern being repeated when dimensions are shared for both logs and metrics:

#[dimension(logs = "server.url.port", metrics = "server.url.port")]

Maybe also allow this syntax?

#[dimension("server.url.port", metrics = true)] // logs included by default