fix(stats): await the async dashboard scope filter (#795)

[SaulMoro]

Follow-up to #795 and #771: main is red since 352cfc4

Summary

#771 and #795 each passed review alone, but together they break teamai stats. #795 changed the scope filter's contract and moved the reported snapshots, and #771's call sites in stats.ts still use the old ones.

 showStats
-  filterEventsByScope(events, { projectRoot } | { excludeProjectRoots })   // sync, old argument
+  await filterEventsByScope(events, config)                               // the call pull makes
   unreportedDashboardStats
-    readJson(~/.teamai/dashboard/reported-*.json)                         // no longer written since #795
+    readReportedInterventions(config) / readReportedPromptTokens(config)  // the scope's own, as pull reads

• Commit 1 makes main green. The user-scope test now follows fix(report): each scope keeps its own reported dashboard snapshots (#786) #795's rule: events with no dataHome are never reported by the user scope, so they are not counted either. fix(report): each scope reports its own dashboard sessions once, against its own snapshots (#785, #786) #791 is still open and changes that rule again (legacy events whose cwd resolves to the user scope), so fix(report): each scope reports its own dashboard sessions once, against its own snapshots (#785, #786) #791 owns flipping that test.
• Commit 2: after the upgrade, stats subtracted a stale shared snapshot, so every session reported since then counted twice. It now reads the scope's own snapshots through team-push's readers.
• fix(stats): count each session once and keep other projects out #771's cwd-based project-root resolution is removed with the old argument.

Evidence

	origin/main	this PR
npx tsc --noEmit	4 errors in src/stats.ts	clean
npx vitest run	11 failed (stats-scope.test.ts: events is not iterable)	303 files, 4640 passed, 1 skipped
new test: own snapshot vs stale shared one	expected 2 to be 1	passes
npm run build	builds (tsup does not type-check)	builds

Real CLI, sandbox HOME, run from the project: project scope (http), team file holds 1 session, events hold sess-1 (reported) and sess-2 (new) in this scope plus sess-3 from another scope, shared snapshot {}, project snapshot has sess-1.

before$ teamai stats --by-repo
TypeError: events is not iterable
    at aggregateSessionMetrics ... at showStats                exit=1

after$ teamai stats --by-repo
  Sessions:           2          # 1 reported + sess-2; sess-3 out, sess-1 once
  Conversation turns: 2
  Tokens (total):     300
By Repo (local event log):
  proj-a  2 sess, 2 turns, 0 tools, 300 tok                    exit=0

Covered one agent-independent path (stats reads local files only); no provider × agent matrix.

Merge Danger

Door: two-way

Local read path only; nothing written to the team repo. A stats run seeds the scope snapshot from the shared file once, the same seed the next pull would write.

Blast Radius: teamai stats

Headline and --by-repo / --by-time for the user scope no longer count events recorded before events carried a data home, matching what pull reports.

[SaulMoro]

@jeff-r2026 heads-up: main has been red since 352cfc4 (#795), and this PR fixes it.

c3bfef3  #771  stats.ts calls the dashboard scope filter synchronously
352cfc4  #795  the filter becomes async (it carried #791's first commit)
         → 4 tsc errors in src/stats.ts, 11 failing tests in stats-scope.test.ts,
           and `teamai stats` crashes with "events is not iterable"

Each PR was green on its own; the break only shows once both are on main. The second commit fixes the other half of the same clash: stats still subtracted the shared reported-*.json that #795 stopped writing, so it would have counted newly reported sessions twice.

The other open #752 PRs, once rebased onto main, show the same stats failures until this one lands, so it would help to merge this one first.

[github-actions]

No findings.

The PR description provides sufficient testing evidence, including a representative real-CLI end-to-end run for the runtime behavior change.