fix(common): use explicit buckets for foyer_hybrid_op_duration

foyer_hybrid_op_duration is the only histogram in Metrics::new registered without explicit buckets, so it falls back to the registry's defaults. For the Prometheus registries those are 5 ms to 10 s, which puts almost every in-memory hit, and many disk hits, into the lowest bucket. Register it with the same 1 µs to 16 s exponential buckets as foyer_storage_inner_op_duration.

Validation:

• cargo test -p foyer-common --lib metrics (13 tests passed)
• cargo clippy -p foyer-common --all-features --lib --tests -- -D warnings
• cargo fmt --all -- --check
• git diff --check