[RadixTree][4/N Refactor]: Move available_and_evictable_str to individual radix cache classes

[pansicheng]

Motivation

Modifications

The available_and_evictable_str method is now an abstract method in BasePrefixCache that each radix cache implementation must provide, rather than a generic function in common.py with isinstance checks.

Accuracy Tests

python3 benchmark/gsm8k/bench_sglang.py --num-shots 8 --num-questions 1319 --parallel 1319

Qwen3-Next-80B-A3B-Instruct mamba
Accuracy: 0.949
Invalid: 0.000
Latency: 26.027 s
Output throughput: 8175.272 token/s

Qwen3-32B full
Accuracy: 0.927
Invalid: 0.000
Latency: 34.917 s
Output throughput: 5637.058 token/s

gpt-oss-120b swa
Accuracy: 0.803
Invalid: 0.043
Latency: 48.075 s
Output throughput: 9391.664 token/s

Benchmarking and Profiling

Checklist

• Format your code according to the Format code with pre-commit.
• Add unit tests according to the Run and add unit tests.
• Update documentation according to Write documentations.
• Provide accuracy and speed benchmark results according to Test the accuracy and Benchmark the speed.
• Follow the SGLang code style guidance.

Review Process

1. Ping Merge Oncalls to start the PR flow. See the PR Merge Process.
2. Get approvals from CODEOWNERS and other reviewers.
3. Trigger CI tests with comments or contact authorized users to do so.
   • /tag-run-ci-label, /rerun-failed-ci, /tag-and-rerun-ci
4. After green CI and required approvals, ask Merge Oncalls to merge.

[gemini-code-assist]

Warning

You have reached your daily quota limit. Please wait up to 24 hours and I will start processing your requests again!

[hzh0425]

/tag-and-rerun-ci

[ispobock]

Could you also test chunk cache (disable radix cache) case?

[pansicheng]

--disable-radix-cache

Qwen3-Next-80B-A3B-Instruct mamba
Accuracy: 0.939
Invalid: 0.000
Latency: 69.834 s
Output throughput: 3050.199 token/s

Qwen3-32B full
Accuracy: 0.927
Invalid: 0.000
Latency: 166.645 s
Output throughput: 1167.150 token/s

gpt-oss-120b swa
Accuracy: 0.800
Invalid: 0.052
Latency: 66.569 s
Output throughput: 6632.504 token/s