docs(blueprints): screen specs, wireframes and component inventory

[Pericena]

Refs #23

What

• docs/blueprints/screens.md: one section per screen ID of docs: MVP user flows #22 (S01–S10), each with:
  • its purpose and flows;
  • the data shown, named as in docs/domain/model.md;
  • the loading, empty, error and success states;
  • the actions and where they lead;
  • the components it uses.
• Component inventory: every atom, molecule and organism in puls3_flutter/lib/src/ui, with the screens that use it.
• docs/blueprints/wireframes/: 32 PNGs.
  • Desktop (1440 px) and mobile (390 px) for every screen.
  • Loading, empty and error wireframes for S02 and S03.
  • Rendered by puls3_flutter/tool/wireframes, which lives outside flutter test, so CI does not run it: built screens use the app with fake data; S05, S06 and S10 are drawn from the same components after F6/F7.
  • The test font draws text as blocks, so they read as low-to-mid fidelity wireframes.
• flows.md links the specs and marks S09 as built (feat: wallet connection in Flutter app #25).

Wireframes: the PNGs in docs/blueprints/wireframes/ . @TOMOKI977, let me know if you still need one.

How to verify

• Compare the screen IDs in flows.md with docs/blueprints/wireframes/: every ID has -1440 and -390.
• For S02, S03 and S07, check the 4 states in screens.md and that the field names match docs/domain/model.md.
• Regenerate the PNGs: cd puls3_flutter && flutter test tool/wireframes/wireframes_test.dart --update-goldens

Frontend owner sign-off

As the frontend owner, I confirm these specs are enough to build every screen without design questions.

[XxHugheadxX]

Approved as a partial delivery of #23 (Refs #23).

• screens.md: it covers S01–S10 with purpose, flows, data, the 4 states, actions and components.
• Names:
  • the data fields match docs/domain/model.md, including Hire.runtimeStatus and Hire.failureReason;
  • S05 and S06 use the ADR-0005 states and the approve, reject and refund actions from the updated flows (#140);
  • a reject from open reads "Cancelled".
• Wireframes: every screen ID has -1440 and -390 PNGs, and they come from the real components.

Before #23 can close:

• The Figma link is still an acceptance criterion of #23 and isn't here. Either add it, or agree on the issue to drop it in favour of the generated PNGs.
• Unreadable labels in the wireframes. The test font (Ahem) draws every glyph as a block, so no label can be read. For example, S06 doesn't show which button is Approve and which is Reject. Loading the app's real fonts in tool/wireframes before rendering would make them readable, and it fits "low to mid fidelity". The goldens can do that with FontLoader and the bundled font files.
• Missing state wireframes. The other screens (S04–S10) document their loading, empty and error states in text only. That's allowed ("its own wireframe or a clear note"), but for S06 the states matter most, so one wireframe for its failed-run and approval-deadline states would help.

[cloudflare-workers-and-pages]

Deploying puls3 with    Cloudflare Pages

Latest commit:	16472d4
Status:	✅  Deploy successful!
Preview URL:	https://728b0963.puls3-4lw.pages.dev
Branch Preview URL:	https://docs-23-screen-specs.puls3-4lw.pages.dev

View logs

[Pericena]

Thanks for the review! Done in 16472d4:

• Readable labels: the wireframe tool now loads Roboto and the Material icons from the Flutter SDK (and the app's Doto), so every label reads (S06 shows Approve and Reject and refund).
• S06 states: new failed-run and deadline-passed wireframes at 1440 and 390 px, linked from its states in screens.md.

On Figma: I'll ask on #23 whether the generated PNGs can replace it, since they're regenerated from the app with one command and stay in sync with the code.