Correct offset in the weights reader (tiny YOLO v2 Coco)

[ameroyer]

When trying to load the tiny YOLO v2 configuration and weights (trained on COCO) from the pjreddie website I got an AssertionError: expect 44948596 bytes, found 44948600. (Using the tiny-yolo.cfg from this repository returned an Over-read tiny-yolo.weights error)
In the parser in the darknet github, the seen metadata is saved as size_t type rather than int which seemed to mess with the initial offset.

[codecov-io]

Codecov Report

Merging #642 into master will decrease coverage by 0.04%.
The diff coverage is 60%.

@@            Coverage Diff             @@
##           master     #642      +/-   ##
==========================================
- Coverage   66.82%   66.77%   -0.05%     
==========================================
  Files          33       33              
  Lines        2101     2104       +3     
==========================================
+ Hits         1404     1405       +1     
- Misses        697      699       +2

Impacted Files	Coverage Δ
darkflow/utils/loader.py	83.8% <60%> (-1.49%)	⬇️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 5fe906b...f47ca50. Read the comment docs.