Search Recorded Metrics by Name, Tags, and Recency

Implement recording and searching for an in-memory collection of metrics. A metric has a name, timestamp, integer value, and string-to-string tags. A search supplies a name, required tags, and a limit `n`, and returns the most recent matching metrics. For this practice version, one callable first records the supplied metrics in order and then answers each search against the completed collection. ### Input - `metrics`: an array of records represented as `[name, timestamp, value, tags]`, where `tags` is a string-to-string map. - `searches`: an array of queries represented as `[name, tags, n]`. ### Output Return one array of matching metric records for each query, preserving query order. Each returned record has the same four-field representation as the input. ### Matching and Ordering Rules - Names must match exactly and are case-sensitive. - For this practice version, every query tag must be present with the same value in a metric's tags. Additional metric tags are allowed. An empty query-tag map imposes no tag restriction. - “Most recent” means greatest timestamp, regardless of recording order. - For equal timestamps, return the metric recorded later first. - Keep duplicate records as separate recorded metrics. Do not collapse them by name, timestamp, or content. - Return at most `n` matches; fewer are valid when fewer exist. If `n == 0`, return an empty array. ### Constraints and Edge Cases - For this practice version, `0 <= metrics.length <= 10000` and `1 <= searches.length <= 500`. - Names and tag keys are nonempty ASCII strings; tag values may be empty strings. - Timestamps are integers from `0` to `1000000000`, and metric values are integers from `-1000000000` to `1000000000`. - Each tag map contains at most ten key-value pairs, and `0 <= n <= metrics.length`. - The total number of returned records across all queries is at most `20000`. - A query with no matching name or tags returns an empty array. ### Example 1 ```text metrics = [ ["cpu",10,1,{"host":"a","env":"prod"}], ["cpu",20,2,{"host":"a"}], ["cpu",20,3,{"host":"b"}], ["memory",30,4,{"host":"a"}] ] searches = [["cpu",{"host":"a"},2], ["cpu",{},2]] output = [ [["cpu",20,2,{"host":"a"}], ["cpu",10,1,{"host":"a","env":"prod"}]], [["cpu",20,3,{"host":"b"}], ["cpu",20,2,{"host":"a"}]] ] ``` The first search permits additional metric tags. The second uses recording order to break the timestamp tie. ### Example 2 ```text metrics = [["latency",5,7,{}]] searches = [["cpu",{},1], ["latency",{},0]] output = [[], []] ``` One query has no matching name; the other explicitly requests zero results.