feature - add execution observations and adapter coverage checks (#66, #67)

[dannymeijer]

Summary

This PR adds RFC 032 execution observations for Session.execute, Session.collect, and Session.write while preserving the existing Result[...] APIs. It also adds the first RFC 033 adapter coverage surface: explicit adapter requirements, conservative coverage records, and Session.check_coverage(...) for requirements supplied by callers.

Type of change

• Bug fix
• New feature
• Refactor / maintenance
• Documentation
• CI / tooling
• RFC (adds/updates docs/rfcs/*)

Area(s)

Select the primary areas touched (labels sync from checked lines when the triage workflow runs):

• Package & tests
• Specification (RFCs)
• Documentation
• Automation & repo config
• Other

Key details

• User-facing behavior: Session now exposes execute_observed, collect_observed, and write_observed methods that return observed result carriers with data, observation, and error fields. Observations include plan and execution-attempt targets, client-session context, backend name, status, diagnostics, wall-clock and monotonic timing, optional row/byte counts, adapter/profile placeholders, trace IDs, and linked coverage records.
• Internals: Observation construction uses the same execution path as the existing execute, collect, and write methods, so legacy callers keep the same Result[...] behavior. Prism plan IDs are centralized through prism_plan_id(...) so inspection and execution evidence do not diverge. Adapter coverage is conservative: unknown stays unknown, DataFusion only reports the capability families this slice can justify, and unsupported backend capability is not modeled as Substrait semantics.
• Risks: This adds public evidence models and observed result carriers. The RFC 033 coverage map is intentionally narrow and explicit-requirement only; automatic requirement inference from inspection remains future work.

Testing / verification

• make ci (or make fmt-check, make build, make test)
• Manual verification described below

Manual verification notes:

• make pre-commit INCAN=/Users/danny/Development/encero/tmp/incan-v03-verify/target/debug/incan passed with Incan 0.3.0: fmt-check, test-style, vocab companion tests, registry metadata, build, and all 256 InQL tests.
• /Users/danny/Development/encero/tmp/incan-v03-verify/target/debug/incan test tests/test_execution_observations.incn passed: 7 focused tests.
• git diff --check HEAD^..HEAD passed.
• Local verification used a temporary ignored incan symlink to mirror CI's sibling Incan checkout required by vocab_companion; the symlink was removed afterwards.

Docs impact

• No docs changes needed
• Docs updated
• Docs follow Divio intent (tutorial/how-to/reference/explanation) where applicable

If docs updated:

• Link(s): docs/language/reference/execution_context.md, docs/language/reference/inspection.md, docs/release_notes/v0_1.md, docs/rfcs/032_execution_observations.md, docs/rfcs/033_adapter_requirements_coverage.md, docs/rfcs/README.md

Checklist

• I kept public docs user-focused and moved internals to contributing docs when appropriate
• I avoided duplicating canonical install/run instructions in multiple places
• I added/updated tests where it materially reduces regressions

Refs #66
Refs #67