Issue: is:escalating filter appears to misclassify issues during short-term spikes

### Problem Statement

The `is:escalating` issue filter does not appear to align with intuitive or expected behavior during short‑term error spikes. Issues returned by this filter are often not meaningfully “escalating,” even when there is a clear spike in errors over a shorter time window.

This makes it difficult to rely on the filter for operational triage during incidents

### Solution Brainstorm

One of the following would be expected:

- `is:escalating` should correlate with a recent increase in frequency or impact over an appropriate short-term window (e.g., minutes or hours), or
- The definition of “escalating” should be clarified if it is intentionally comparing broad timeframes (e.g., week-over-week).

It appears that escalation may be determined by comparing this week vs. last week, which may be too coarse for operational use during real-time incidents. e.g. an alert reviewing the last 15 mins showed a solid bump in count for issues, but none of the issues listed in the Alert showed as "escalating", even though one clearly has an increase in count.

使用涵蓋最近 15 分鐘的警示重現該行為，然後將它顯示的 issue 與 `is:escalating` 的結果以及篩選器目前的比較時間範圍進行比較。完成的標準是：要麼讓篩選器與短期頻率或影響的增加保持一致，要麼記錄它刻意使用的較寬時間範圍。

由索引模型根據 Issue 內容生成。