epic: improve high-volume inbox triage workflows

[cboone]

Context

After extended real-world Fastmail cleanup sessions, several recurring pain points emerged around safety, scalability, and operator ergonomics in bulk triage.

This epic tracks the core improvements needed to make jm a first-class triage tool.

Goals

• reduce manual scripting around bulk actions
• make large operations safe and reliable
• improve filter precision and docs discoverability
• standardize triage reporting

Child Issues

• feat: support query-based bulk actions #24 feat: support query-based bulk actions
• feat: add --dry-run mode to action commands #25 feat: add --dry-run mode to action commands
• feat: add exact sender and domain filters #26 feat: add exact sender and domain filters
• fix: make bulk operations resilient to JMAP size limits #27 fix: make bulk operations resilient to JMAP size limits
• docs: align help output and CLI reference for tag filters #28 docs: align help output and CLI reference for tag filters
• feat: add a triage summary command for inbox cleanup #29 feat: add a triage summary command for inbox cleanup

Suggested Delivery Order

1. fix: make bulk operations resilient to JMAP size limits #27 and feat: add --dry-run mode to action commands #25 (safety + reliability)
2. feat: support query-based bulk actions #24 and feat: add exact sender and domain filters #26 (workflow + filter precision)
3. docs: align help output and CLI reference for tag filters #28 (docs parity)
4. feat: add a triage summary command for inbox cleanup #29 (operator UX)