golang concurrency improvements

[tirprox]

1. Using smaller integer size
2. Instead of allocating tagged post count slice each time, allocate one slice per worker thread and zero it with each new task
3. Reducing some other allocations
4. Removed branch in a for loop when counting tagged posts

Overall my results are much closer to rust rayon version now:
Running Go with concurrency
Benchmark 1: ./related_concurrent
Processing time (w/o IO) 6.947594ms
Processing time (w/o IO) 9.377195ms
Processing time (w/o IO) 6.417481ms
Processing time (w/o IO) 7.270459ms
Processing time (w/o IO) 7.082818ms
Processing time (w/o IO) 7.084531ms
Processing time (w/o IO) 7.056418ms
Processing time (w/o IO) 6.957182ms
Processing time (w/o IO) 7.204956ms
Processing time (w/o IO) 7.444545ms
Processing time (w/o IO) 7.835368ms
Processing time (w/o IO) 6.815176ms
Processing time (w/o IO) 6.613258ms
Time (mean ± σ): 26.2 ms ± 0.9 ms [User: 93.0 ms, System: 11.2 ms]
Range (min … max): 25.1 ms … 27.6 ms 10 runs

Checking output
Output file: related_posts_go_con.json is valid!

[jinyus]

Thanks for the PR, great work!