Fix kv pool evicting more tokens than needed

[Edenzzzz]

Motivation

Following #4913, seems that we only need to evict tokens_required - avail_size(). cc @xiezhq-hermann

Modifications

Checklist

• Format your code according to the Code Formatting with Pre-Commit.
• Add unit tests as outlined in the Running Unit Tests.
• Update documentation / docstrings / example tutorials as needed, according to Writing Documentation.
• Provide throughput / latency benchmark results and accuracy evaluation results as needed, according to Benchmark and Profiling and Accuracy Results.
• For reviewers: If you haven't made any contributions to this PR and are only assisting with merging the main branch, please remove yourself as a co-author when merging the PR.
• Please feel free to join our Slack channel at https://slack.sglang.ai to discuss your PR.

[xiezhq-hermann]

technically it is true, but the eviction has always been kind of generous to prevent too frequent eviction, i.e., not to evict if not necessary, and to evict sufficient once it happens. I have no objection to this change but just think it might be not necessary.
if we are making this change, would mind examine eviction of other places to align with this style?

[Edenzzzz]

I see, I think you may be right, collecting, heapifying and iterating through the leaves incurs some overhead, so we should evict generously