Investigate migrating from unmaintained tsup to tsdown

[cliffhall]

Background

The CLI and TUI clients (and test-servers) currently bundle with tsup, which is effectively unmaintained. tsdown is the actively-maintained, Rolldown-powered successor positioned as a drop-in replacement.

Goal

Investigate why we're on tsup and evaluate migrating to tsdown.

Tasks

• Inventory every place tsup is used (clients/cli, clients/tui, test-servers, any others) and the config/options each relies on.
• Assess tsdown compatibility with those configs (entry points, formats, dts, externals, the @inspector/core alias, watch mode).
• Prototype the migration for one client and confirm builds + tests pass.
• Document findings and recommend whether/how to migrate.

Acceptance criteria

• A clear recommendation on migrating to tsdown, with the migration path and any blockers identified.

[cliffhall]

Triage: Priority = Medium

Scored 6/16 with the priority rubric added to AGENTS.md in #1891. This card is already approved and sitting in Todo — only Priority was set here, its Status is unchanged.

Axis	Score	Reasoning
Severity / impact	2/5	Build tooling hygiene. tsup is effectively unmaintained but nothing is broken; this is an investigation whose outcome may legitimately be 'stay'.
Urgency / staleness	3/5	v2.2.0 milestone. No forcing function beyond the maintenance risk of sitting on an unmaintained bundler.
Signal bonuses	+1	milestone (v2.2.0) (+1)
Total	6	Lands in the Medium band (6-8).

Related issues. One connected issue:

• chore(deps): untangle the vitest/Storybook exact-pin knot to clear GHSA-p63j-vcc4-9vmv #1839 — same theme of dev-dependency maintenance debt, though a different tool and a different reason

Scores are a starting point, not a verdict — the rubric is meant to be overruled when it's plainly wrong, provided the reason is written down.