[feature] 提供设置连续熔断的timeWindow次数

## Issue Description

feature request

### Describe what you expected to happen

连续N次被熔断后，就不希望有请求进来了。

目前sentinel的机制是下次timeWindow会放一个请求进来进行判断（HALF_OPEN），如果失败继续熔断。

对于明确知道对端已经下线的场景，每次timeWindow的检测其实是没必要的。

可以放开一个参数设置，比如连续被熔断的timeWindow次数，来实现这个目的。

I expect sentinel stop allowing new requests coming in when circuit-breaker taking effect in continuous N timeWindow.

The issue does not name files or tests. Start by locating the circuit-breaker implementation and its HALF_OPEN handling, then determine how timeWindow transitions are represented. Done should include a defined configuration for consecutive timeWindow failures and behavior that prevents probe requests for the configured period, with tests covering the transitions.

Written by the indexing model from the issue text.