feat(skills): add code-review skill for GCSFuse

[kislaykishore]

Description

Adds a new, highly refined custom code-review and quality assurance skill runbook under .gemini/skills/code-review/SKILL.md.

This developer utility skill equips AI agents (such as Antigravity) with localized engineering knowledge of:

1. Effective Go & GCSFuse Concurrency Mutexes: Mutual exclusion and read-write locking verification via custom debug wrappers (locker.New & locker.NewRW tracking locks held for >5s), locking boundaries (releasing mutexes prior to remote network or disk calls), and avoiding lock contentions under parallel reads (JAX/PyTorch model checkpoints).
2. Modern Python PEP 8 Guidelines: PEP 8 checking, typing parameters, robust context resource management, exceptional guarding scopes, and mock-patch isolation.
3. Flat Bucket Directory Emulation: Dynamics of emulating virtual parent folder inodes under --implicit-dirs versus explicit directory placeholders, and folder rename consistency (HNS buckets vs flat bucket copying).
4. Append Composing & Garbage Collection: Composition chains utilizing TmpObjectPrefix for sequential write appends exceeding threshold boundaries, and periodic garbage collection sweeps to purge residual temporary files.
5. OTel Telemetry Metric Filtering Gotcha: Critical validation warning that custom metrics must begin with approved prefixes (fs/, gcs/, file_cache/, buffered_read/, grpc., read/), otherwise the OpenTelemetry exporter silently discards them during aggregation.
6. Localized Integration Testing Taxonomy: Maps codebase edits onto specific E2E test suites inside tools/integration_tests/ (e.g. implicit_dir, mounting, read_cache, concurrent_operations, stale_handle, monitoring, readonly).

Link to the issue in case of a bug fix.

N/A

Testing details

1. Manual - Executed unified static compilation validation target make build successfully with exit status 0 (0 issues).
2. Unit tests - N/A
3. Integration tests - N/A

Any backward incompatible change? If so, please explain.

N/A