tailtriage v0.3.0

v0.3.0 is now published on crates.io

Highlights

• New tailtriage-tracing crate for converting tt.* tracing span evidence into standard tailtriage Run artifacts.
• Optional tailtriage facade features for tracing intake integrations.
• JSONL tracing import support for persisted span records.
• Live in-memory tracing recorder APIs for collecting completed tracing spans and converting them into tailtriage runs.
• Optional Tokio session integration for coupling tracing intake with Tokio runtime sampling.
• CLI tracing import command for producing analyzable tailtriage Run JSON from tracing JSONL input.
• Semantic tt.* tracing field convention for request, stage, and queue spans.

Changed

• Expanded the release surface from direct instrumentation-only workflows to include tracing-based intake workflows.
• Kept tracing import output aligned with the existing Run JSON artifact contract and analyzer path rather than introducing a separate tracing-specific analyzer.
• Tightened imported tracing evidence validation around required fields, malformed tt.* spans, duplicate request IDs, child-span correlation, timestamp ordering, and persistable zero-request artifacts.
• Added durable import warnings to Run metadata so conversion-quality issues remain visible during later analysis.

Fixed

• Prevented persisted tracing imports from silently producing analyzer-hostile zero-request Run artifacts.
• Improved handling of tracing spans with missing optional outcome/success fields by defaulting conservatively while surfacing warnings.
• Improved correlation of imported stage and queue spans to retained request intervals, including truncation accounting when matching requests exceed capture limits.