fix(analytics): retain Countme zeros and preserve missing values as gaps

[castrojo]

Fixes #1086

Problem

Project Bluefin family cards on the analytics page computed isTracked = count > 0. A real reported zero (e.g. Dakota's weekly value of 0) was indistinguishable from a missing week, so:

• the sparkline history was cleared to [] entirely instead of rendering with a gap for the missing point, and
• the card fell back to its bootstrapping/provisioning placeholder text instead of showing the actual trend.

Fix

Added a shared readCount() helper that returns the reported value (including 0) and only returns null when a field is genuinely absent or non-finite — matching the gapSafe() convention already used by the comparative time-series charts in this file. isTracked, the per-card count, and the 12-week sparkline history now all key off count !== null rather than count > 0, so:

• a real 0 renders as 0 (not a placeholder), and
• only genuinely missing weeks render as sparkline gaps.

Testing

• npx tsc --noEmit — no new errors
• npx eslint src/components/analytics/CountmeAnalyticsCharts.tsx — clean
• npx prettier --check — clean

— hive: backend=copilot model=claude-sonnet-5

🐝 Hive Agent: contributor | SHA: ed9a243e