Add shared histogram buckets for duration metrics

[sjmiller609]

Summary

• add a shared default duration bucket set for Hypeman duration histograms
• apply the shared buckets across HTTP, instance lifecycle, guest ops, guest memory, volumes, VMM, and egress proxy histograms
• use a slower build-specific bucket set for build and image build duration histograms

Testing

• go test -run '^$' ./lib/middleware ./lib/builds ./lib/otel
• go test -run '^$' ./lib/images ./lib/guest ./lib/volumes ./lib/vmm ./lib/guestmemory ./lib/egressproxy
• go test ./lib/instances -run 'TestSnapshotCompressionMetrics_RecordAndObserve|TestInstanceOldestInStateMetric_ObserveOldestAgePerState|TestInstanceTimeToRunningMetric_RecordWhenBootMarkersPersisted|TestLifecycleDurationMetrics_RecordCompressionLabels'

---

Note

Medium Risk
Changes histogram bucket boundaries for multiple exported duration metrics, which can affect Prometheus/OTel time series, dashboards, and alert thresholds even though runtime behavior is otherwise unchanged.

Overview
Introduces shared OpenTelemetry duration histogram bucket helpers in lib/otel/buckets.go, including a common bucket set and a slower build-specific set.

Updates duration histograms across builds/images, HTTP middleware, instances lifecycle/snapshot operations, guest ops, guest memory, volumes, VMM, and egress proxy to use these shared bucket boundaries (removing egress-proxy’s local bucket definitions).

^{Written by Cursor Bugbot for commit efd6256. This will update automatically on new commits. Configure here.}