Optimizations

[GeorgCantor]

Changes Made:

Single-pass processing - Replaced multiple collection operations with a single iteration over input files
Set-based deduplication - Used linkedSetOf() instead of distinct() for O(1) uniqueness checking

Before:

Created intermediate collection gavs
3 separate map operations + 3 distinct() calls
Total complexity: ~O(6n) operations

After:

Processes each file once during iteration
Immediate deduplication via Set insertion
Total complexity: O(n) with better constant factors
Lower memory usage (no intermediate collection)

Result: More efficient code with better memory and time performance characteristics, especially for large input sets.

[martinbonnin]

I'm sorry but given the size of the datasets and this running in a much more complex process (Gradle), the gain is probably negligible. From a cursory look, the code looks correct but I just don't think the change is worth it.

I can change my mind if you can share real life numbers.