基于ConcurrentLinkedHashMap实现的CacheMap会导致在高CPU核数下，单个接口产生的ParameterMetric对象很大的问题

## Issue Description

ConcurrentLinkedHashMap在构造二维数组时会根据CPU核心数来计算出当前数组的一维下标长度，当核数为96时，下标为128，此时单个ParameterMetric对象大小接近1M，如下图所示：


当接口调用量较大时，此处将会占用大量内存，sentinel是否有考虑过这块的内存使用，或者切成其他Cache？毕竟googlecode.concurrentlinkedhashmap作者也建议使用Caffeine了.

### Tell us your environment
springboot 2.3.10
sentinel 1.7.2
springcloudgateway 2.2.6

### Anything else we need to know?
sentinel在后续版本中是否有意向对这块进行修改？

Start by locating CacheMap and its ConcurrentLinkedHashMap usage, then inspect how ParameterMetric objects are created and retained under high CPU-core counts. The issue provides no file or test entry point; done would require an agreed cache design, a reproducible memory comparison, and regression coverage.

Written by the indexing model from the issue text.