v0.3.0

Added

A client for the v2 API, under client.v2 — 407 operations across 90
resources. Everything you use today is untouched: v2 is a new namespace beside
the existing surface, not a replacement for it.

There are two ways in, and they compose:

// Flat path — each segment consumes exactly one URL path id
await client.v2.workspaces.projects.states.list("acme", "ENG");

// Loaded rows — a fetched row carries its ids, so nothing repeats
const project = await client.v2.workspaces.projects.retrieve("acme", "ENG");
await project.cycles.list();
await project.workItems.create({ name: "Ship it" });

Path ids are named after the resource they identify, singular, with no Id
suffix: retrieve(slug, project, state).

• Navigable rows — 19 Loaded* types. A fetched workspace reaches 25
  children, a project 19, a work item its comments, links, attachments and
  worklogs.
• Compile-time field projection — fields narrows the return type through
  Pick<T, F | "id">, so reading a field you did not request is a type error
  caught before you run, not undefined at runtime. expand is exposed wherever
  the API declares it.
• Typed filters and ordering — every query filter the API declares is
  reachable and typed; order_by likewise, as generated unions.
• Pagination — both envelopes: offset (per_page, offset) and cursor
  (cursor). iterate() follows pages on either, so the difference stays out of
  your code.
• Lookup by name — findByName(...), and findByDisplayName(...) on
  property resources, so scripts can use the names people see in the UI instead
  of ids.
• Bridges — add / remove on relationship collections, e.g.
  cycle.workItems.add([...]), up to 100 ids per call.
• Bulk writes — bulkCreate / bulkUpdate / bulkDelete where the API
  offers them, up to 50 items per call.

v2 types are exported from the package root, including the Loaded* row types.

Full Changelog: v0.2.14...v0.3.0