[v1] Refactor KVCacheManager for more hash input than token ids

[rickyyx]

This PR refactors the KVCacheManager such that:

1. It will be easier to pass in more metadata for hashing (e.g. lora id, mm_hash and etc)
2. Remove the needed of tracking token ids inside a block (which was used for hash calculation, but no longer needed since hash is assigned to a block).

Benchmarking

Running the below on main and the PR shows no diff on performance.

VLLM_USE_V1=1 python3 benchmarks/benchmark_prefix_caching.py \
--model meta-llama/Meta-Llama-3-8B-Instruct  \
--num-prompts 50 \
--input-length-range 500:500  \
--prefix-len 250     --seed 0 --output-len 500

[github-actions]

👋 Hi! Thank you for contributing to the vLLM project.
Just a reminder: PRs would not trigger full CI run by default. Instead, it would only run fastcheck CI which starts running only a small and essential subset of CI tests to quickly catch errors. You can run other CI tests on top of those by going to your fastcheck build on Buildkite UI (linked in the PR checks section) and unblock them. If you do not have permission to unblock, ping simon-mo or khluu to add you in our Buildkite org.

Once the PR is approved and ready to go, your PR reviewer(s) can run CI to test the changes comprehensively before merging.

To run CI, PR reviewers can do one of these:

• Add ready label to the PR
• Enable auto-merge.

🚀

[comaniac]

LGTM. Great job!