fix: resolve infinite loop in path search algorithm

[Waleed-Ahmad-dev]

What does this PR do?

This PR fixes a bug where the graph traversal algorithm gets caught in an infinite loop during complex format conversions (like pptx to pdf via pandoc).

Fixes #605

Why is it needed?

Because pandoc registers a massive number of input/output combinations (creating over 4,000 edges), the pathfinding algorithm was experiencing an exponential path explosion. The custom visitedBorder logic was failing to properly prune redundant paths, causing the browser/app to hang indefinitely.

How was it fixed?

• Replaced the global visited array and visitedBorder logic with a minCosts Map.
• Implemented standard Dijkstra pruning: the algorithm now records the lowest cost to reach any specific format node.
• If a newly discovered path to a node costs more than or equal to a previously found path, it immediately discards it, preventing infinite loops and drastically speeding up the search.
• Removed the now-obsolete visitedBorder property from the QueueNode interface.

How was it tested?

• Tested locally by attempting a conversion from pptx to pdf.
• Verified that the algorithm now resolves the route almost instantly without hanging or crashing the app.