[Conductor] conductor: tiered prefix indexer

[Misak2333]

Description

Third slice of the Mooncake Conductor upstreaming series, on top of the
prefix index types and hash strategy. Adds PrefixCacheTable, the
in-memory index that answers cache-hit queries across GPU, CPU and disk
tiers:

• prefixindex/prefix_indexer.{h,cpp}: engine registration and profile
  binding validation, the GPU and shared-tier store/remove/clear
  mutations, the Query prefix walk, and GetGlobalView. Blocks are keyed
  by ProjectedPrefix and carry per-tier owner sets, so one walk reports
  the longest match per instance plus per-dp-rank and per-tier hit
  counts, with optional cache_salt and instance filtering.
• Two-level locking: a global context-map mutex guards the context
  lookup, and each ContextState has its own mutex. Contexts are held by
  shared_ptr so a query drops the map lock before touching state, and
  the lock order is fixed one way (map, then state) to keep it
  deadlock-free. Query does not reorder entries, so read-only traffic
  keeps a shared lock.
• Capacity control: per-context block limit (200k by default, zero
  disables it) with insertion-order eviction that drops the oldest
  entries in batches down to 90% occupancy, and a cumulative eviction
  counter that surfaces stored/removed events going out of sync.
• prefix_indexer_test_peer.h exposes the internals the tests need
  (presence snapshots, order-metadata sizes, state locks) without
  widening the public surface, through a single friend declaration.

Module

• Transfer Engine (mooncake-transfer-engine)
• Mooncake Store (mooncake-store)
• Reshard (mooncake-reshard)
• Mooncake EP (mooncake-ep)
• Mooncake PG (mooncake-pg)
• Integration (mooncake-integration)
• P2P Store (mooncake-p2p-store)
• Python Wheel (mooncake-wheel)
• Common (mooncake-common)
• Mooncake RL (mooncake-rl)
• CI/CD
• Docs
• Other

Type of Change

• Bug fix
• New feature
• Refactor
• Breaking change
• Documentation update
• Performance improvement
• Other

How Has This Been Tested?

Test commands:

# Example: bash scripts/run_ci_test.sh

Test results:

• Unit tests pass
• Integration tests pass (if applicable)
• Manual testing done (describe below)

Checklist

• I have performed a self-review of my own code
• I have formatted my code using ./scripts/code_format.sh
• I have run pre-commit on the files changed in this PR and all hooks pass
• I have updated the documentation (if applicable)
• I have added tests to prove my changes are effective
• For changes >500 LOC: I have filed an RFC issue

AI Assistance Disclosure

• No AI tools were used
• AI tools were used (specify below)