Scenario

A logistics company runs a nationwide package-routing network with millions of daily scans. Each scan can introduce a “link” between two hubs (undirected). Over time, redundant links create cycles, which makes certain downstream processes (like deterministic route replay and auditing) harder.

Your job is to prune the network into a spanning forest (a tree per connected component) by removing edges that are redundant.

Problem

You are given:

• An integer n (number of hubs labeled 0..n-1)
• A list edges where each element is [u, v] representing an undirected link between hubs u and v

Return a list of edges to remove such that:

1. If you process edges in the given order, you keep an edge if it connects two previously disconnected components.
2. Otherwise, the edge would create a cycle (or is a parallel duplicate within the same component), so you remove it.

Output requirements

• Return the removed edges in the exact orientation they appear in the input (do not reorder endpoints).
• Return removed edges in the order they are removed during the left-to-right scan.

This produces a deterministic spanning forest consistent with the ingestion order.

Examples

Example 1

• Input: n = 5, edges = [[0,1],[1,2],[2,0],[1,3],[3,4]]
• Output: [[2,0]]
• Explanation: After keeping [0,1] and [1,2], edge [2,0] connects nodes already connected, so it’s removed.

Example 2

• Input: n = 6, edges = [[0,1],[1,2],[2,0],[3,4],[4,5],[5,3],[0,2]]
• Output: [[2,0],[5,3],[0,2]]
• Explanation: [2,0] closes a cycle in component {0,1,2}; [5,3] closes a cycle in {3,4,5}; [0,2] is a parallel redundant edge and is removed when encountered.

Notes

• The graph may be disconnected.
• Parallel edges may exist.
• If the graph is already a forest, return [].