fix(usage): clarify Cursor pools and small usage shares

[rynfar]

Small nonzero usage shares appear as 0.0%, and Cursor Limits currently shows the combined monthly total beside its two allowances without explaining their relationship. This displays a precision floor for positive shares, labels and explains the Cursor pools on web and mobile, and hides the combined card when every account it represents supplies both pools. Combined-only accounts from older hosts and unfamiliar windows stay visible. Existing account navigation, reset controls and source identity remain intact.

This ports independent presentation hunks from frozen T3 source e5a46d6c5d00b89afba5274a94d42428c8d79763 (#10409). It does not change the maintainer's Cursor account-history deferral: no dashboard history, Keychain setting, credential collection, provider refresh or contract version changes. Local OpenCode history remains covered by #886 and Pylon's Antigravity reader is retained. Generic complete/partial scan merging and source diagnostics from the same source remain separate audit follow-ups.

Verification: 150 focused tests across shared formatting/pooling/merge, Cursor limits, and the real web/mobile component entry points; typechecks for shared, server, web and mobile; scoped lint/format and diff checks; production web build and production iOS/Android exports. The formatter and both client regressions fail on the original implementation. Tests also cover mixed-version accounts, unknown windows, immutable inputs, mobile account navigation and changing environment selection.

Draft: required before/after UI images remain outstanding. Browser/Simulator computer-use permission is pending; no rendered browser/native pass, live-account validation, release or installation is claimed.

Follow-up to upstream catch-up #865. Implemented and reviewed by GPT-6 in the Codex harness.

---

^{Need help on this PR? Tag @codesmith-bot with what you need. Autofix is disabled.}

Evidence limit and review

Rendered check on an isolated dev server with real Claude/Codex accounts: the Usage page renders unchanged for non-Cursor providers. This host has no Cursor account and no sub-0.1% shares, so the Cursor pool labels and <0.1% formatting could not be shown live; they are covered by the focused web/mobile tests that fail on base. Screenshots of the account usage page were not posted because this repository is public.

Independent adversarial review: MERGE. git merge-tree is clean against #954/#955; recommended order #953 → #954 → #955. Note recorded: the web pooled-limits view now hides the account name and refill line for single-account bars for every provider (matches upstream; mobile still shows the refill line).

Review by Claude Opus 5.5 in Claude Code.

[github-actions]

Thread transfer impact

✅ Thread transfer remains within every enforced ceiling.

Provider	Metric	Main baseline	This PR	Impact	PR ceiling
Codex	Total thread wire	14.0 KiB	14.0 KiB	−2 B (−0.0%)	15.1 KiB	✅
Codex	Thread snapshot wire	7.3 KiB	7.3 KiB	+1 B (+0.0%)	7.3 KiB	✅
Codex	Live turn WebSocket wire	6.7 KiB	6.7 KiB	−3 B (−0.0%)	7.8 KiB	✅
Codex	Live turn WebSocket decoded	58.0 KiB	58.0 KiB	0 B (0.0%)	66.4 KiB	✅
Codex	Live turn messages	9	9	0 (0.0%)	21	✅
Claude	Total thread wire	14.0 KiB	14.0 KiB	+14 B (+0.1%)	15.1 KiB	✅
Claude	Thread snapshot wire	7.3 KiB	7.3 KiB	−19 B (−0.3%)	7.3 KiB	✅
Claude	Live turn WebSocket wire	6.7 KiB	6.8 KiB	+33 B (+0.5%)	7.8 KiB	✅
Claude	Live turn WebSocket decoded	58.9 KiB	58.9 KiB	+44 B (+0.1%)	66.4 KiB	✅
Claude	Live turn messages	9	10	+1 (+11.1%)	21	✅

Baseline: b5a607e · PR result: c97a744 · Source CI: success

Scenario and decoded snapshot size

10 historical turns, 5 command tools per turn, 878.9 KiB retained MCP result per historical turn, and a 1.05 MiB retained result in the measured turn.

• Codex decoded thread snapshot: 115.7 KiB
• Claude decoded thread snapshot: 116.4 KiB

Updated in place by a trusted workflow. PR artifacts are strictly validated and never executed.