[BUG] Make indexing stop status reflect what actually stopped #36

Type

Bug

Problem / Value

Stopping indexing now immediately reports an idle state and a generic stopped message. The same message can appear when nothing is running, and it no longer explains that a partial index was preserved.

Context

Deferred finding 🟡4 from PR #34: #34, which addresses #23. Source reviewed: 8c38f1c.

The stop path no longer enters the stopping state or uses the partial-index message. Those state definitions, guards, and translated messages remain. This user-interface change was not covered by the design. Qualification: an existing parser failure is preserved by a guard; the generic status is not unconditional in that case.

Reproduction

Proposed regression cases: stop during a scan with completed work, stop while watching for file changes, stop when already idle, and stop after a parser failure.

Acceptance criteria

• Agree and document the intended stop-status behavior before changing it.
• Show a stopping state while shutdown is pending if that behavior is retained.
• Explain retained partial progress when appropriate; do not imply work was stopped when nothing was running.
• Preserve parser-failure messages.
• Align remaining state definitions, guards, and translations with the chosen behavior.
• Add tests for the listed stop cases.

Constraints/Preferences

Keep stopping prompt and preserve indexed data. Do not silently introduce another user-interface behavior change.