feat(monitoring): wire AwsCostCollectorService into cost scheduler

[sadiqabubakar826-del]

Wire real AWS Cost Explorer data into the hourly cost scheduler

Previously, CostSchedulerService published a hard-coded 0 as the hourly cost metric on every
cycle, making any capacity planning or cost alerting built on that metric reason from a
constant rather than real data. AwsCostCollectorService already existed and could fetch live
Cost Explorer data, but was never connected to the scheduler.

───────────────────────────────────────────────────────────────────────────────────────────────

Changes

AwsCostCollectorService

• Removed the duplicate @Cron decorator and CostTrackingService injection — the service is now
  a pure data-fetcher.
• collectHourlyCost() returns HourlyCostResult | null ({ amount, billingPeriod }) instead of
  writing the metric itself. Returns null on any failure or when the collector is disabled, so
  the caller can decide how to handle it.
• billingPeriod is formatted as YYYY-MM-DD/YYYY-MM-DD, matching Cost Explorer's TimePeriod
  convention and making the reporting delay explicit.

CostTrackingService

• recordHourlyCost(amount, billingPeriod?) now accepts an optional billing period.
• The infrastructure_hourly_cost_usd gauge carries a billing_period label so dashboards and
  alert rules can see which window a data point describes.

CostSchedulerService

• Injects AwsCostCollectorService and drives the collection cycle.
• On success: records the real amount and billing period via CostTrackingService.
• On failure (null result): leaves the previous gauge value unchanged and increments the
  cost_collection_failures_total counter — failures are visible without publishing a fabricated
  value.
• No placeholder, no TODO.

cost-scheduler.service.spec.ts (new)

• 8 tests covering: delegation to the collector, amount and billing period forwarding, failure
  counter incremented on each failed cycle, no metric write on failure, and no hard-coded
  placeholder published.

───────────────────────────────────────────────────────────────────────────────────────────────

Testing

pnpm test --testPathPattern="cost-scheduler" # 8/8 pass
pnpm test --testPathPattern="monitoring" # 61/61 pass (no regressions)
pnpm typecheck # clean

───────────────────────────────────────────────────────────────────────────────────────────────

Acceptance checklist

• Cost metric sourced from Cost Explorer, not a constant
• Collector failure leaves previous value and increments cost_collection_failures_total
• Metric labelled with the billing period it describes
• No placeholder or TODO remains in the scheduler

closes #1037

[drips-wave]

@sadiqabubakar826-del Great news! 🎉 Based on an automated assessment of this PR, the linked Wave issue(s) no longer count against your application limits.

You can now already apply to more issues while waiting for a review of this PR. Keep up the great work! 🚀

Learn more about application limits

[RUKAYAT-CODER]

Well done on the job done so far!
Kindly fix workflow to pass.