Return Top K Relevant Apps

You are given a mapping from application name to keyword relevance scores, for example: ```json { "github": {"diff": 0.3, "code review": 0.8}, "jira": {"ticket": 0.7, "workflow": 0.4} } ``` Define an application's overall relevance as the sum of all its keyword scores. Given this structure and an integer `k`, return the names of the `k` most relevant applications sorted by descending relevance. If two applications have the same total relevance, break ties by application name in ascending lexicographic order. Discuss an efficient approach and its time complexity.